Solar Energy 158 (2017) 161–185

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Review

A comprehensive study on different types of faults and detection techniques MARK

for solar photovoltaic system
⁎
Siva Ramakrishna Madeti , S.N. Singh
Alternate Hydro Energy Centre, Indian Institute of Technology Roorkee, Uttarakhand 247667, India

A R T I C L E I N F O A B S T R A C T

Keywords: Monitoring systems are essential to maintain optimal performance of photovoltaic (PV) systems. A critical aspect
Photovoltaic systems in such monitoring systems is the fault diagnosis technique being used. The role of a fault detection and diagnosis
Different faults technique is to identify the causes affecting the real-time energy production and/or smooth functioning of PV
Fault detection techniques systems. Over the past decade, various fault detection methods were reported in literature. Among all the fault
detection techniques reported, some paid significant attention only on faults that occur in the PV system, some
on faults on DC side of the PV system while the rest focused on AC side faults. For the first time, this paper
provides a comprehensive review of popular fault detection techniques, addressing all major types of faults in PV
systems. Detailed insights of PV fault detection techniques along with their relative performances are covered. A
new fault detection technique is also proposed to identify the type and location (module level) of a fault. This
review enables the reader to get acquaintance with major aspects/considerations in developing/choosing an
effective yet viable fault detection technique for small and medium scale PV systems.

1. Introduction estimating the true potential (power) of RES and for fault identification
in the installed area in real-time. Moreover, continuous monitoring of
Owing to the increase in industrial expansion and continuous PV system(s) health are very crucial to detect the causes, which hamper
growth in energy consumption, global energy demand per capita is on the desired performance. A comprehensive solution for all these pro-
rise. This paved the way for intensive research of new, safe and sus- blems is being termed as monitoring system along with fault diagnosis
tainable green power technologies such as, solar, wind, hydro, tidal, techniques, whose job is to maximize the operational reliability of PV
biomass, and geothermal. The primary focus of renewable energy system with minimum system costs and to detect the causes affecting
technologies research is to convert renewable resources into electrical the performance of the PV system.
energy in order to feed the utility grid or consumer loads. Among all the A fault detection algorithm for PV system can provide an accurate
renewable energy sources (RES), solar energy generation is recognized estimation of electricity generation under normal operating condition
as the best way to extract energy from the environment (Madeti and and detection of faults present in the PV system. This would enable
Singh, 2017a,b). Global installed PV capacity at the end of 2016 was operator to take corrective actions, in order to prevent the PV system to
reported as 310 GWp (Moreno-Garcia et al., 2016). However, as they under-perform for prolonged periods of time, which improve the per-
are installed in outdoor environment, continuous exposure to harsh formance of PV system by minimizing the power losses caused by the
environmental conditions (sun beam, rainfall, etc.) may reduce the faults. Over the past decade, different aspects of fault detection tech-
optimal performance of system. A monitoring study was conducted on a niques for PV system were reported. This paper presents an overall
test PV system in Firth et al. (2010) and it was reported that the annual review of existing fault detection techniques in both DC and AC side of
power loss due to various faults is about 18.9%. In addition to this, PV systems and also proposes a low-cost multisensor smart monitoring
there will be a production loss associated with the occurrence of each of system (SMS) along with a fault detection procedure that is intended to
grid fault. All such issues pertaining to different domains have to be detect various faults in PV system (modular level). A comparison of the
addressed by a single solution. Such an endeavor requires the complete present study with the other significant studies as suggested is also
knowledge of meteorological data (climatic conditions, which affects provided in Table 1.
energy production) of the area where the system will be installed. This paper is organized as follows: Section 2 provides an overview
Consequently, it is necessary to develop techniques, which help in of PV monitoring system with fault detection and diagnosis structure.

⁎
Corresponding author.
E-mail address: [email protected] (S.R. Madeti).

http://dx.doi.org/10.1016/j.solener.2017.08.069
Received 22 June 2017; Received in revised form 14 August 2017; Accepted 23 August 2017
0038-092X/ © 2017 Elsevier Ltd. All rights reserved.
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

Table 1
Highlights of some of relevant articles.

S. no. Characteristics of previous Fault classification Fault detection Cost Comments Ref
monitoring system techniques analysis

DC side AC side DC side AC side

1 – ✓ – – – – • Based on location and structure faults in DC side of PV

system is classified
Sabbaghpur Arani
and Hejazi (2016)
• No algorithm provided for fault detection
2 ✓ – – – – – • Overview on pv monitoring system is provided Madeti and Singh
• No faults classification and algorithm provided for
fault detection
(2017a)

3 – – – ✓ – – • Proposed algorithm can only give the possible fault

location and cannot identify the possible fault types
Chine et al. (2014)

4 – ✓ – ✓ – – • No faults classification on AC side and algorithm

provided for fault detection
Sharma and Chandel
(2013)
5 – – – – ✓ – • No algorithm provided for fault detection Li et al. (2014)
6 ✓ ✓ ✓ ✓ ✓ ✓ • Overview on PV monitoring system, classification of
possible faults in DC and AC side of PV systems and
Present study

various fault detection techniques used are provided

• Proposed algorithm can give the possible fault
location and identify the possible fault types

Classification of faults in DC and AC side of PV system is given in connected inverter. There are two categories of PV monitoring systems,
Section 3. Sections 4, the different types of fault detection techniques namely ground based and space based systems. The advantage of
are discussed. Section 5 and 6 addresses the fault detection techniques former approach is that it can respond very quickly and accurately
in DC and AC side of the PV systems. Proposed fault detection technique monitor the health of PV system. Thus, it provides a chance to enhance
has been described in section 7. The results of different fault detection the performance of PV system by detecting possible energy losses from
techniques are discussed in Section 8. Section 9 summarizes the con- changes in operating conditions and/or faults, before they have a
clusion of this work. considerable effect on energy production and/or system health. Sensors
are used as major components in ground based system, which can
measure the parameters in real-time. Though the space based systems
2. Overview of PV monitoring system with fault detection and
are more economical due to absence of sensors. However, there accu-
diagnosis structure
racy is poor and they also get easily affected by the climatic conditions.
Since, these problems are undesirable; therefore, a ground based PV
The general block diagram of PV monitoring system with fault de-
monitoring system has been reported in this work.
tection and diagnosis blocks is shown Fig. 1. It consists of PV array, DC/
Another important block is the fault detection and diagnosis block.
DC converter with maximum power point tracking (MPPT), grid

Fig. 1. General block diagram of PV monitoring system with fault detection and diagnosis blocks.

162
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

Table 2 If the fault is located in DC side of PV system then microscopic

Parameters to be measured. analysis can be conducted to understand the factors responsible for it.
Analytical techniques have been found very useful for microscopic
PV system type Parameters
analysis. The most significant and recent techniques used in micro-
Electrical Meteorological analysis, reported in literature are: (i) Scanning electron microscopy
(SEM) (ii) Attenuated total reflectance infrared microscopy (ATIR) (iii)
Grid connected Utility grid: Photovoltaic (i) Total irradiance,
X-ray micro-tomography. The first indication that PV system has fault is
array: in the plane of
(i) Grid voltage array GT provided by the reduction in DC power output. However, after ana-
(ii) Current to (i) Output (ii) Ambient lyzing the electrical characteristics of the PV system it is confirmed that
utility grid voltage temperature there is fault present in the system. After the fault location is identified,
(iii) Current from (ii) Output (iii) Module temperature the modules are subjected to visual inspection. During visual inspection
utility grid current (iv) Wind speed
(iv) Power to (iii) Output (v) Wind direction
the module is checked for various defects such as yellowing, cracks in
utility grid power (vi) Humidity cell, corrosion in junction box connections, open or short circuit faults,
(v) Power from (iv) Output (vii) Barometric pressure by-pass diode fault, back sheet polymer cracks, bubbles formation and
utility grid energy delamination in the encapsulant, oxidation and discoloration in junc-
(vi) Utility grid
tion cables, encapsulant browning etc. under the 1000 lux light source
impedance
Stand alone Load: as per standards. If the fault occurred in AC side of PV system the power
flow in the system will become zero. Various fault detection techniques
(i) Output voltage for power electronics and islanding detection have been explored in
(ii) Output current literature. However, up to best knowledge of the authors, no work has
(iii) Output power
been reported towards a generalized fault detection technique for
components and islanding detection that is suitable for an arbitrary
If there is fault present in the system, the performance monitor block switching power converters and configuration of PV systems. The next
will set to high state, which in turn activates the fault detection and step is to locate the exact position of the fault in DC and AC side for
diagnosis block. According to the set of fault detection and diagnosis which several techniques have been proposed in literature. The initial
rules in the knowledge –base, the performance of PV system will be documentation of these techniques was reported in Hund and King
analysed and if there is a real fault it can be detected and diagnosed. (1995) and is briefly explained in the following sections.
The type and location of fault present in the system can be identified by
analyzing the measured parameters. These parameters are selected
according to British Standard BS IEC 61724 (International 5. Fault detection techniques in DC side of PV systems
Electrotechnical Commission, 1998). Depending on the type of PV
system, i.e. grid connected or stand-alone type, the parameters to be 5.1. Based on electrical characterization
measured are shown in Table 2. Note that the meteorological and PV
array parameters to be measured are common for both types. 5.1.1. Climatic data independent (CDI) technique
This type of technique does not require any measurement of climatic
data (solar irradiance, temperature, humidity and wind speed) for fault
3. Typical faults in PV systems detection and classification (FDC). This technique use external devices
such as LCR [inductance (L) capacitance (C), and resistance (R)] meters
The array power is equal to the sum of the output power of the and signal generators for fault detection. In order to detect fault, the
individual PV modules. During normal operation the array power is response of the PV system is analyzed right after injection of each
very close to predicted power. However, in field conditions, a number signal. For example, earth capacitance measurement (ECM) is proposed
of factors can cause a PV array to reduce its output power. Any factor by Takashima et al. (2006) to detect disconnection of PV module in a
which reduces the output is considered as “fault”. It may be temporary string. The time-domain reflectometry (TDR) technique is developed by
(shadows, bird droppings, and dust or snow accumulation on the sur- Schirone, measure the electrical characteristics of transmission line to
face of the PV panel) or permanent (electrical disconnection, wiring detect breakdown point, faults and impedance change due to de-
losses and ageing). A permanent fault would remain for prolonged time gradation without requiring climatic data (Schirone et al., 1994). A
whereas a temporary fault can be cleared within a specific time period. statistical approach based on non-parametric Kruskale Wallis (KW) and
Faults can occur in a PV array and generate different effects on the the Analysis of Variance (ANOVA) tests are investigated in Vergura
performance and lifetime of PV system. Therefore, it is necessary to et al. (2008).
identify what kind of failures can be found in the real system. The first
step in this challenge is recognition and classification of all possible
faults in the PV systems. The classification of the most common types of 5.1.2. Electrical current–voltage (I-V) measurement (EM) technique
fault in PV system is presented in Fig. 2 and the definition of faults is Fault detection based on this technique requires only measurement
given in Table 3. of electrical signals from the output terminal, such as voltage and
current (Lin et al., 2017). Microcontroller based real-time expert system
4. Different types of fault detection techniques to monitor, supervise and control PV systems was developed by
Benghanem and Maafi (1998) and one of its subsequent uses was
Recently, efforts are made to understand the failure of PV systems analyzed in Mukaro and Tinarwo (2008). In Chine et al. (2014), auto-
which results in the development of new techniques to detect and locate matic fault detection in grid connected photovoltaic (GCPV) system was
the type of fault present in the system. These techniques have helped in developed by using voltage and current indicators. In Miwa et al.
improving the system reliability and lifetime of PV systems. The clas- (2006), (− dI ) −V is used to detect the partial shading phenomenon.
dV
sification of different fault detection techniques to identify the type and Kaplanis and Kaplani (2011) have calculated fill factor (FF), shunt re-
location of the fault occurring in DC and AC sides of PV system is il- sistance (Rsh), A and Rs from I(V) characteristics to determine the
lustrated in Fig. 3. The comprehensive review of possible faults in PV performance of the system.
system discusses in this paper along with the fault detection techniques
are summarized in Table 4.

163
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

1. Hot spot
1. Partial shading 2. Soldering 1. Shunted by-pass diode
2. Snow covering 3. Degradation of modules Open circuit Open circuit
Upper Lower 2. Inverted by-pass diode 1. Line-to-line
3. Dust/bird dropping/leaves 4. Glass breakage with by-pass without by- Series arc Parallel arc Inverter Lighting
ground ground 3. Short circuited by- 2. Line-to-ground
4. Uniform irradiance 5. Delamination diode pass diode
pass diode
distribution 6. Discoloration
7. Interconnect breakage
of busbars
8. defects in frame
9. cell breakage
10. micro cracks
11. different degradation
rate in between the
modules

Fig. 2. Classification of faults in DC and AC side of PV system.

5.1.3. Comparison between measured and modeled PV system outputs layered ANN, decision tree-based method and graph-based semi-su-
(CMM) technique pervised for FDC of PV plants has been described in Ducange et al.
In this type of technique, the difference between measured and (2011) and Bonsignore et al. (2014). Some other examples for FDC of
modeled PV system outputs is used for fault detection (Chouder and PV systems can be found in Ducange et al. (2011), Chao et al. (2010),
Silvestre, 2010; Silvestre et al., 2013). In order to determine the ex- Karatepe and Hiyama (2011), Cheng et al. (2011), Zhao et al. (2012)
pected output power from the PV system, various predictive models and Zhao et al. (2015).
have been used. Usually, these techniques can establish theoretical
thresholds, which the change in systems output power does not exceed. 5.1.6. Heat exchange and temperature (HET) based models
However, if it does exceed the threshold then the system is considered When the fault occurs in PV array, the temperature of the PV
to be faulty. In some approaches, the expected value of PV outputs was module changes. In order to detect faults, the heat exchange and
predicted with parametric models and weather variables along with module temperature during the faulted condition is used in this tech-
adjustable parameters (Drews et al., 2007). In Chao et al. (2008), the nique. For example, the authors in Hu et al. (2013) and Vergura et al.
authors developed a method based on the extended correlation function (2012) have been modeled the physical defects of different types of PV
and on the matter element model to detect faults in PV array. cells using finite element method. It is based on the thermal behavior of
the PV cells resulting from electrical faults.
5.1.4. Power loss analysis (PLA) technique
This technique is based on the analysis of power losses in the PV 5.1.7. Ground fault detection and interruption (GFDI) fuse
system. For FDC, power losses are calculated by comparing the mon- The path for the intentional circuiting ground current can be pro-
itored data with simulated results. In Chouder and Silvestre (2010), a vided by system grounding during fault condition. When the ground
supervision and fault detection method has been proposed based on current is higher than the safe threshold current limit, then the fuse gets
power losses analyses. The method proposed in Madeti and Singh melted. If the fuse is opened, then inverter should turn OFF im-
(2017), detect faults occurring in both the PV array and in the inverter. mediately to isolate the PV array from the rest of the power system
In Silvestre et al. (2013), captured losses in a PV system, current and (Hernandez and Vidal, 2009; Hernández et al., 2010). In order to avoid
voltage ratios are used for the fault detection algorithm. false tripping due to leakage current the rating of the fuse need to be
high. However, the sensitivity of the GFDI should be set by considering
5.1.5. Machine learning (ML) techniques the array leakage current. The effect of different parameters such as
This technique relies on learning of particular conditions corre- relative humidity, resistance of the grounding conductor, ambient
sponding to a particular type of data. Any newly introduced condition temperature, electromagnetic interference, and salt mist on the leakage
can be identified on the basis of previous training (Schapire, 2008), current have been investigated in the literature (Flicker and Johnson,
which can be used for FDC. Since, the PV output changes under dif- 2013a,b).
ferent climatic conditions; defining theoretical boundaries for fault
detection is a difficult task. Training a model for certain input-output 5.1.8. Residual current monitoring devices (RCDs)
patterns can overcome this limitation. Various machine learning tech- The difference between the current entering and leaving PV system
niques have been employed in the literature for FDC. To identify faults through the positive and negative current carrying conductors (CCCs)
in PV array due to partial shading and inverter, expert systems were can be sensed by these devices. To protect the PV system against L-L
used in Yagi et al. (2003). In Li et al. (2012), artificial neural network and L-G faults, RCDs can be installed for each string or for entire array
(ANN) has been used to classify different faults in PV array. Takagie (Brooks, 2012). By considering the leakage current of PV modules the
Sugeno Kahn Fuzzy Rule (TSKFRBS), Bayesian belief networks, three- sensitivity of a RCD should be set. In Hernandez and Vidal (2009), it is

164
 Table 3
Definition of Faults in DC and AC side of PV system.
S.R. Madeti, S.N. Singh

Side Type of fault Description Ref

DC side MPPT fault Sensor disconnection to the converter installed with MPPT or problem in MPPT charge controllers Jiang and Maskell (2015)
Mismatch fault Temporary • Partial shading • Partial shading caused by trees/buildings, overhead power lines Nguyen (2015), Hu et al. (2015), Heidari et al. (2015)
• Snow covering • Depending on the weather conditions and geographical location and Sabbaghpur Arani and Hejazi (2016)
• Dust/bird dropping/leaves • Dust/bird dropping/leaves accumulation on the surface of PV module
• Uniform irradiance • Different solar radiation intensity during the day
distribution
Permanent • Hot spot • Degrade the optical and mechanical properties of encapsulation material due to poor thermal Sharma and Chandel (2013) and Köntges et al. (2014)
• Soldering conductivity of cell encapsulation (EVA) materials
• Degradation of modules • During manual soldering the cell breakage rate is high
• Glass breakage • Due to corrosion and electro-migration in interconnects and contact layers. The interconnect
• Delamination degradation affects both shunt and series resistance whereas contact degradation can increase the
• Discoloration series resistance
• Interconnect breakage of • Caused by poor packaging during transportation, installation or by hail and stone throwing etc.,
busbars The moisture intrusion in PV modules deployed in the field show that moisture penetration results
in frame
• into delamination and reduces the active area of the module
• Defects
breakage to low quality EVS used, the additive in EVA partly disappears and due to UV radiation and
• Cell • Due
heat the EVA starts to get yellow or brown color
• Micro cracks changes in temperature are known to wear out the cell interconnections
• Frequent
to heavy snow load on PV modules installed at heavy snow regions
• Due to big ribbon with a big solder bond, it puts more local stress on the cell and causes them to be

165
• Due
more likely to break the cell
• Wafer slicing, cell production, stringing and other embedded process during the production of the
solar cell and module can cause cell cracks
Short circuit fault (any module or by-pass diode) Bad connection between the solar cells or manufacturing defects Jiang and Maskell (2015) and Wolgemuth (2010)
Inverted bypass diode fault Incorrect connections made by the operator Chine et al. (2015)
Bypass diode fault
Open circuit with bypass diode fault Bad connection between solar cells; broken cells; hot spots; loose connections; aged power cables near
Open circuit with-out bypass diode fault terminals; plugging and unplugging connectors at junction boxes leads to open circuit in PV module
Line to line (L-L) fault It occurs when unintentional connection between two nodes of the PV array Alam et al. (2015) and Zhao (2010)
Line to ground (L-G) fault It occurs when one conductor drops to the ground or comes in contact with the neutral conductor
Bridging fault Connection resistance between PV modules or cabling Davarifar et al. (2013)
Ground fault upper ground fault An unintentional low impedance path between the current carrying conductors (CCC) of the last two Alam et al. (2015)
modules in a PV string and the ground
lower ground fault An unintentional low impedance path between the CCC of the second and third modules in a PV string
and the ground with large back feed current
Arc fault Series arc fault Series arc fault occurs when there is discontinuity in any of CCC resulting from damage in cell,
corrosion of connectors, solder disjoints, etc.
Parallel arc fault It occurs mostly due to insulation breakdown.

AC side Inverter failure High thermal and mechanical stress on power switches can provoke a faulty operation of the Chan and Calleja (2006)
converters. Also, the failure of other components such as capacitors, drive circuitry can also cause the
inverter failure
Grid failure Due to line tripping, equipment failure, human error, disconnection for maintenance services, or Zhao (2010)
network reconfiguration
Sudden natural disasters Total blackout due to lightning, storm, and so forth Omer (2007)
Solar Energy 158 (2017) 161–185
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

Climatic data independent technique

Electrical current–voltage measurement technique

ISLANDING
DETECTION
Comparison between measured and modeled PV system outputs technique

Power loss analysis technique

Machine learning techniques

Analog Digital
Heat exchange and temperature based models implementation implementation

Ground fault detection and interruption fuse

Time domain Frequency domain
Residual current monitoring devices

Insulation monitoring devices

Frequency spectrum analysis of the voltage or current waveforms

Estimating randomness in the voltage signal

Spread spectrum time-domain reflectometry

Fig. 3. Classification of different fault detection techniques.

recommended that the set point of differential current (ΔI) at which an contain less arcing energy and frequency above 500 kHz interacts with
RCD signals a ground faults should be chosen according to the following external RF noise source. Therefore, these frequency levels are not used
equation: in arc-fault circuit interrupters (AFCIs ) . Unfortunately, most inverters,
charge controllers and DC/DC converters have switching frequencies
ΔI ⩾ Csl × Ileak,max (1)
within the range of 10–50 kHz. These converters and charge controller
where Ileak,max is the maximum leakage current and Csl is the multiplier. also generate harmonics and sub-harmonics. Therefore, based on one
The values of Csl must be >1 to avoid false tripping due to measurement frequency component it is difficult to develop arc fault detection
error or external noise. technique. The most suitable frequency range for PV plants is about
1–100 kHz (Johnson et al., 2011; Johnson and Kang, 2012; Seo et al.,
5.1.9. Insulation monitoring devices (IMDs) 2012). The detector developed in Haeberlin and Real (2007), contains a
These devices measure the resistance between CCCs and ground. If resonant circuit which is tuned to a specific frequency. This frequency is
the resistance is lower than the preset value (Rf thre ) then the device can not expected during normal operation of the plant. By monitoring this
alarm the system. IMDs are used to detect the ground fault in a frequency, an arc can be discovered.
grounded system by disconnecting the GFDI fuse at the time of mea-
surement. Since the insulation resistance is influenced by the ambient 5.1.11. Estimating randomness in the voltage signal (ERV)
conditions, the threshold value for a false tripping (Rf thre ) is suggested This technique for arc fault detection uses filters and statistical
in Ball et al. (2013) and is given as follows: methods in order to analyze the measured signal. This can be done in
the time domain using digital Finite impulse response band pass
Rf thre ⩽ Csr × Riso min (2)
(FIR) filters and some simple calculations. The output is a signal that
where Riso min is the minimum insulation resistance and Csr is the mul- represents the level of chaos in the input signal. This level can be
tiplier, similar to Csl in (1) . The value of Csr should be <1 to avoid false compared with a threshold to get the information if an arc exists or not.
tripping. In Schimpf and Norum (2009), by using analog and digital approach the
authors have analyzed the measured signals.
5.1.10. Frequency spectrum analysis (FSA) of the voltage or current
waveforms 5.1.12. Spread spectrum time-domain reflectometry (SSTDR)
This technique analyzes the lower frequency content to detect the SSTDR meter uses pseudorandom binary noise modulated high-
arc faults since the frequency content of an arc shows a 1/ f relation. frequency sine wave to generate an autocorrelation plot. By using the
However, due to signal variations from irradiance level, partial shading, incident and reflected signals this plot can be generated. The fault
or noise from the inverter the frequencies below 1000 Hz are not re- present in the system is detected by using autocorrelation plot. SSTDR
commended to avoid false tripping. Arcing frequency above 100 kHz can be used without disconnecting the inverter. In Alam et al. (2014),

166
 Table 4
Analysis of fault detection and classification methods on DC and AC sides of the PV system.

Author Ref. Type of PV Parameters Sampling Monitoring Types of faults considered Fault detection methods Software/ Validation
system period method programming
Meteorological. Electrical. DC side AC side language Simulated Experimental
S.R. Madeti, S.N. Singh

Stellbogen Stellbo (1993) Stand-alone Gg, Gd, Tm Varr, Iarr, Vmp, Imp 1hr PC ✓ – CMM PVNode & Shade ✓ –
Schirone Schirone et al. Stand-alone – Vstr 5 μs PC ✓ – CDI – – ✓
(1994)
Hund Hund and King Stand-alone/ T a, H Ultrasonic scan of PV – – ✓ – Ultrasonic inspection – – ✓
(1995) Grid tied module method
Bower Bower and Wiles Stand-alone/ – Ileak, Ibackfeed, Vbackfeed 2s – ✓ – GFDI – – ✓
(2000) Grid tied
Task Task (2002) Grid tied – Vgrid, f and Zgrid – PC – ✓ HI & CGI MATLAB ✓ ✓
Yagi Yagi et al. (2003) Grid tied Gg, Tm, Ta Varr, Parr, Vac, Pac 1h PC ✓ ✓ ML NEDO ✓ –
Hung Hung et al. (2003) Grid tied – Iac, F, Vgrid 10 ms PC – ✓ Active technique (active MATLAB ✓ ✓
frequency drift)
Guasch Guasch et al. Stand-alone Gg, Tm, Tbat, TL Varr, Iarr, Vbat, Ibat, VL, 1h PC ✓ – CMM MATLAB ✓ ✓
(2003) IL,
Jang Jang and Kim RDGs – F, Vgrid, ϕ, THD, 30 ms PC – ✓ Passive technique PSCAD/EMTDC ✓ –
(2004) (harmonics measurement
and voltage unbalance)
Yu Yu et al. (2005) Grid tied – Vgrid, f – PC – ✓ ARPV PSIM ✓ ✓
Yin Yin (2005) RDGs – F, Vgrid 0.1 s PC – ✓ Passive technique – ✓ –
(harmonics
measurements)
Miwa Miwa et al. (2006) Stand-alone Gg, Tm Varr, Iarr 5 min PC ✓ – EM – – ✓
Takashima Takashima et al. Stand-alone – Vstr 5 ns – ✓ – CDI – – ✓
(2006)

167
Ropp Ropp et al. (2006) Grid tied – Ipcc, Iarr 5 ms PC – ✓ Active technique MATLAB ✓ ✓
(impedance
measurement)
Drews Drews et al. (2007 Grid tied Gg, GdGt, Tm, Ta Parr, Pac 1h PC ✓ – PLA MATLAB ✓ ✓
Menon Menon and Nehrir RDGs – Vgrid, F, Fset point, Iac 5s PC – ✓ Hybrid technique MATLAB ✓ –
(2007) (Voltage unbalance and
frequency set point)
Ezzt Ezzt et al. (2007) RDGs – df 1s PC – ✓ Hybrid technique MATLAB ✓ –
Φ,
dt (Voltage fluctuation
injection)
Pigazo Pigazo et al. RDGs – Vac, Iac, Pac, Vgrid, PL 0.2 s PC – ✓ Signal processing MATLAB ✓ –
(2007) technique (Discrete
Wavelet transform)
Vergura Vergura et al. Stand- alone – PV energy 10 min PC ✓ – CDI MATLAB ✓ –
(2008)
Chao Chao et al. (2008) Stand- alone Gg, Tm Varr, Iarr, Parr – PC ✓ – CMM PSIM ✓ –
Moradzadeh Moradzadeh et al. RDGs – Voltage unbalance 0.08 s PC – ✓ Hybrid technique MATLAB ✓ –
(2008) (voltage unbalance and
frequency set point)
Takashima Takashima et al. Stand- alone – Vstr 5 μs – ✓ – CDI – – ✓
(2008)
Karimi Karimi et al. RDGs – Vpcc, Igrid, Ppcc, Pgrid, 0.02 s PC – ✓ Active technique (Sandia PSCAD/ EMTDC ✓ –
(2008) PL, F voltage shift)
Zhu Zhu et al. (2008) RDGs – Vac, Iac, Pac, Vgrid 30 ms PC – ✓ Signal processing MATLAB ✓ –
technique (Continuous
Wavelet transform)
Schimpf Schimpf and Stand-alone/ – Vmod, Imod, IL 50 μs – ✓ – ERV – – ✓
Norum (2009) Grid tied
(continued on next page)
Solar Energy 158 (2017) 161–185
 Table 4 (continued)

Author Ref. Type of PV Parameters Sampling Monitoring Types of faults considered Fault detection methods Software/ Validation
system period method programming
Meteorological. Electrical. DC side AC side language Simulated Experimental

Takashima Takashima et al. Grid tied – Vstr 5 μs – ✓ – CDI – – ✓

S.R. Madeti, S.N. Singh

(2009)
Liu Liu et al. (2010) Grid tied – Vpcc, Fpcc, Varr, Iarr 0.04 s PC – ✓ Active technique (slip MATLAB ✓ ✓
mode frequency shift)
Quintana Quintana et al. Stand-alone/ Ta Ultrasonic scan of PV – – ✓ – CT scan/Ultrasonic – – ✓
(2009) Grid tied module inspection method
Wu Wu et al. (2009) Grid tied Varr, Iarr, Vg – PC ✓ ✓ ML MATLAB ✓ –
Mahat Mahat et al. RDGs – F, Vgrid 0.1 s PC – ✓ Hybrid technique DigSILET ✓ –
(2009) (Technique based on PowerFactory
voltage and real power 13.2.334
shift)
Hernandez Hernandez and Stand-alone/ T a, H Rs_ins, Rp_ins, Ileak, Voc, – – ✓ – IMD Classical approach – ✓
Vidal (2009) Grid tied V l,
Pigazo Pigazo et al. RDGs – F, Vac, Iac, Pac, Vgrid, PL 0.2 s PC – ✓ Signal processing MATLAB ✓ ✓
(2007) technique (Discrete
Wavelet transform)
Lezana Lezana et al. – – Vo, IL 1 ms PC – ✓ Model-based (signal MATLAB ✓ ✓
(2009) (FDI in processing in frequency
power domain)
converters)
Li Li et al. (2009) RDGs – Vpcc, Iac, Vac 0.2 s PC – ✓ Active technique PSCAD/ EMTDC ✓ ✓
Sleszynski Sleszynski et al. – – Phase currents, 0.1 s PC – ✓ Model-free (based on MATLAB ✓ ✓
(2009 diagnostic signal (FDI in current analysis)
power

168
converters)
Zeineldin Zeineldin and RDGs – Vpcc, F 1s PC – ✓ Passive technique PSCAD/EMTDC ✓ –
Kirtley (2009) (Under/over voltage and
under/over frequency)
Lee Lee and Park RDGs – Vpcc, Ipcc, F 10 ms PC – ✓ Passive technique PSCAD/EMTDC ✓ –
(2010)
Hu Hu and Sun RDGs – Vac, F, Vgrid 0.05 s PC – ✓ Passive technique MATLAB ✓ –
(2009) (voltage phase jump)
Samantaray Samantaray et al. RDGs – Vac, Iac, Pac, Vgrid, PL 5 ms PC – ✓ Signal processing MATLAB ✓ –
(2009) technique (Discrete
Wavelet transform)
Firth Firth et al. (2010) Grid tied Gg, Tm, Ta Parr, Pac, Energy to and 5 min – ✓ ✓ PLA – – ✓
from grid
Chouder Chouder and Grid tied Gg, Tm Varr, Parr, Vac, Pac 1 min PC ✓ – PLA MATLAB ✓ ✓
Silvestre (2010)
Antionio Rodríguez-Blanco – – Gate signal, Rg, Vge, 10 μs PC – ✓ Model-free MATLAB ✓ ✓
et al. (2011) Vge*, Vt (FDI in
power
converters)
Chao Chao et al. (2010) Grid tied Gg, Tm Vmp, Imp, Pmp and Voc – PC ✓ – ML Solar Pro ✓ –
Hanif Hanif et al. (2010) Grid tied – Iac, Igrid, IL, Vpcc, F, 0.1 s PC – ✓ Signal processing MATLAB ✓ –
technique (Discrete
Wavelet transform)
Aljankawey Aljankawey et al. RDGs – Vgrid, Igrid, F, earth – – – ✓ Passive technique – – –
(2010) reference (voltage to (Voltage phase jump
earth, impedance to detection)
earth), Pac, Qac
Houssein Houssein et al. Stand- alone Gg, Tm Vmp, Imp and Pmp 1 min PC ✓ – CMM MATLAB ✓ ✓
(2010)
(continued on next page)
Solar Energy 158 (2017) 161–185
 Table 4 (continued)

Author Ref. Type of PV Parameters Sampling Monitoring Types of faults considered Fault detection methods Software/ Validation
system period method programming
Meteorological. Electrical. DC side AC side language Simulated Experimental

Morsi Morsi et al. RDGs – ROCOP 2 ms PC – ✓ Signal processing MATLAB ✓ –

S.R. Madeti, S.N. Singh

(2010) technique (Wavelet

packet transform)
Hernandez Hernández et al. Stand-alone/ Gg, Ta, H, Tm Rs_ins, Ileak – – ✓ – GFDI Classical approach – ✓
(2010) Grid tied
Polo Polo et al. (2010) Grid tied Gg, Tm & Sw Vmp, Imp 1h PC ✓ – PLA – ✓ ✓
Shariatinasab Shariatinasab and RDGs – Igrid, Iac 25 s PC – ✓ Signal processing ETAP/MATLAB ✓ –
Akbari (2010) technique (Discrete
Wavelet transform)
Amaral Amaral and – Tcap ESR, Xcap, Rl – PC – ✓ Model-based MATLAB ✓ ✓
Cardoso (2012) (FDI in
power
converters)
Fayyad Fayyad and RDGs – Vgrid, Pac, Qac 0.016 s PC – ✓ Signal processing PSCAD/EMTDC ✓ –
Osman (2010) (WT[ANN classified])
Ray Ray et al. (2010b) Grid tied – Vac, Vgrid – PC – ✓ Signal processing MATLAB ✓ –
technique (S-transform)
Zhao Zhao et al. (2011) Grid tied Gg, Tm Varr, Iarr, Il-l – PC ✓ – EM MATLAB ✓ –
Dini Dini et al. (2011 Stand-alone/ – Varc, Iarc, Vstr 0.1 s – ✓ – FSA – – ✓
Grid tied
Ducange Ducange et al. Stand-alone Gg, Tm Varr, Iarr, Parr 1h PC ✓ – ML MATLAB ✓ ✓
(2011)
Ketola Ketola and Norris Stand-alone/ – Parr, Rs, Isc, Voc, EL – – ✓ – EL – – ✓
(2011) Grid tied imaging of PV module

169
Karatepe Karatepe and Stand-alone Gg, Tm Vmp, Imp – PC ✓ – ML – ✓ –
Hiyama (2011)
Bahrani Bahrani et al. RDGs – Vpcc, Iac, F 0.02 s PC – ✓ Active technique (Sandia PSCAD/ EMTDC ✓ –
(2011) voltage shift)
Wang Wang et al. Stand-alone/ T a, H Rcont, Ileak – – ✓ – GFDI – – ✓
(2011) Grid tied
Xu Xu et al. (2011) Stand-alone Gg, Tm Varr, Iarr, Vmod, Imod 2h PC ✓ – EM – ✓ –
Yu Yu et al. (2011) RDGs – Vac, F 0.5 s PC – ✓ Passive technique MATLAB ✓ ✓
Yazdani Yazdani et al. – – Vdc_link 2 ms PC – ✓ Model-based PSCAD/EMDTC ✓ ✓
(2011) (FDI in
power
converters)
Johnson Johnson et al. Stand-alone/ – Line Frequency, Vnoise – – ✓ – FSA – – ✓
(2011) Grid tied
Zeineldin Zeineldin and RDGs – Vgrid, F, Pac, Qac, NDZ 1s PC – ✓ Active technique (sandia PSCAD/ EMTDC ✓ –
Salama (2011) phase shift)
Cheng Cheng et al. Stand-alone Gg, Tm Varr, Iarr – PC ✓ – ML – ✓ ✓
(2011)
Veldman Veldman et al. Stand-alone/ – Lock in – – ✓ – LIT/EL/ Ultrasonic – – ✓
(2011) Grid tied thermography/EL inspection
imaging/X-ray/
ultrasonic scan of PV
module
Zhao Zhao et al. (2011) Grid tied Gg, Tm Varr, Iarr 1 min PC ✓ – CMM MATLAB ✓ ✓
Yoo Yoo et al. (2011) Grid tied – F, ϕ, Vgrid 1s PC – ✓ Active technique (active PSIM ✓ –
phase shift)
Johnson Johnson et al. Stand-alone/ Gg, Ta, H F, Vnoise, Inoise, Vstr, Istr 0.1 s ✓ – FSA Sandia model – ✓
(2011) Grid tied
(continued on next page)
Solar Energy 158 (2017) 161–185
 Table 4 (continued)

Author Ref. Type of PV Parameters Sampling Monitoring Types of faults considered Fault detection methods Software/ Validation
system period method programming
Meteorological. Electrical. DC side AC side language Simulated Experimental

Duan Duan et al. (2011) – – Phase currents, – PC – ✓ Digital implementations MATLAB ✓ ✓

S.R. Madeti, S.N. Singh

(FDI in (based on analysis of the

power characteristic of current
converters) signals)
Ray Ray et al. (2011) Grid tied – Vac, Vgrid – PC – ✓ Signal processing MATLAB ✓ –
technique (S-transform)
ElNozahy ElNozahy et al. RDGs – Vgrid, F 0.02 s PC – ✓ Intelligent classifiers MATLAB ✓ –
(2011) (Decision tree)
Samantaray Samantaray et al. RDGs – ROCOF, ROCOP, Vgrid 0.2 s PC – ✓ Intelligent classifiers MATLAB ✓ –
(2011) (Probabilistic Neural
Network)
Drif Drif et al. (2012) Grid tied Gh, Gt, Ta Varr, Iarr 1h PC ✓ – CMM MATLAB ✓ ✓
Yafaoui Yafaoui et al. Grid tied – Iac, Vac, IL, VL, F 2 ms PC – ✓ Active technique (active MATLAB ✓ ✓
(2012) frequency drift with
positive feedback)
Gokmen Gokmen et al. Grid tied Gg, Tm Pstr 1h PC ✓ – CMM PSIM ✓ ✓
(2012)
Texas instruments Instruments Stand-alone/ – Varc, Iarc, Vstr, Istr 0.1 s – ✓ – FSA – – ✓
(2012) Grid tied
Li Li et al. (2012) Stand-alone Gg, Tm Vmp, Imp – PC ✓ – ML MATLAB ✓ –
Zhao Zhao et al. (2012) Grid tied Gg, Tm Varr, Iarr 5s PC ✓ – ML WEKA ✓ ✓
Flicker Flicker and Stand-alone/ Gg, Ta Rs_ins, Zarr, Ileak, Ifault, – PC ✓ ✓ RCD Sandia model ✓ ✓
Johnson (2013b) Grid tied Iac, Voc, Isc
Braun Braun et al. Stand-alone Gg, Tm Varr, Iarr, Vmod, Imod – PC ✓ – CMM MATLAB ✓ –

170
(2012)
Lin Lin et al. (2012) Stand-alone Gg, Tm Vmod, Imod, Vbat, Ibat – Online ✓ – CMM MATLAB ✓ ✓
Ciampi Ciampi et al. Grid tied Gg, Tm Varr, Iarr 1h PC ✓ – ML SUNinspector ✓ ✓
(2012)
Vergura Vergura et al. Stand-alone/ Tm, Ta, Sw Varr, Iarr – PC ✓ – HET/Thermography – – ✓
(2012) Grid tied
Karegar Karegar and RDGs – Vac, Vgrid, F 0.2 s PC – ✓ Signal processing MATLAB ✓ ✓
Sobhani (2012) technique (Discrete
Wavelet transform)
Gokmen Gokmen et al. Stand-alone Gg, Ta Vstr – PC ✓ – CMM – ✓ ✓
(2013)
Wheeler Shao et al. (2013) Stand-alone/ – Iarm_conv, Vcell_cap 0.1 s PC – ✓ Model-based (sliding MATLAB ✓ –
Grid tied (FDI in mode observer)
power
converters)
Guerriero Guerriero et al. Grid tied – Vmod, Imod, Voc, Isc 3 ms – ✓ ✓ CDI & EM – – ✓
(2013)
Pu Pu et al. (2013) – – ESR, Vc, Vdc_link, VESR 0.1 s PC – ✓ Model-based MATLAB ✓ ✓
(FDI in
power
converters)
Ball Ball et al. (2013) Stand-alone/ Gg Rs_ins, Zarr, Ileak, Ifault, – PC ✓ ✓ IMD Sandia ✓ ✓
Grid tied Iac, Voc, Isc
Silvestre Silvestre et al. Grid tied Gh, Gt, Ta Varr, Iarr, Vac, Iac 5 min PC ✓ ✓ PLA LabVIEW ✓ ✓
(2013)
Shahbazi Shahbazi et al. – – Vac, Iac, Iphase 100 μs PC – ✓ Model-based (signal MATLAB ✓ ✓
(2013) (FDI in processing in frequency
power domain)
converters)
(continued on next page)
Solar Energy 158 (2017) 161–185
 Table 4 (continued)

Author Ref. Type of PV Parameters Sampling Monitoring Types of faults considered Fault detection methods Software/ Validation
system period method programming
Meteorological. Electrical. DC side AC side language Simulated Experimental

Flicker Flicker and Stand-alone/ Gg, Ta Ileak, Rleak, Zarr, IGFDP, GFDI SPICE ✓ ✓
S.R. Madeti, S.N. Singh

Johnson (2013c) Grid tied Varr, Iarr, Parr

Haney Haney and Stand-alone/ Gg, Ta Varr, Iarr, Parr, IR – PC ✓ – Visual inspection – – ✓
Burstein (2013) Grid tied imaging of a PV
module, Rinsu, Voc, Imp
Hu Hu et al. (2013) Stand-alone Gg, Ta, Sw Vmod, Imod – PC ✓ – CMM & Thermal camera MATLAB ✓ ✓
Lin Lin et al. (2013) – – Phase currents – PC – ✓ Model-free MATLAB ✓ ✓
(FDI in
power
converters)
Youssef Youssef et al. – – Vgrid, Iac, Vac, Pac – PC – ✓ Model-based (observer MATLAB ✓ ✓
(2013) (FDI in based)
power
converters)
Zhao Zhao et al. (2013 Grid tied – Varr, Iarr, Istr 1s PC ✓ – CDI MATLAB ✓ ✓
Ribeiro Ribeiro et al. Stand alone – Varr, Iarr, Vc1, Vc2, Vac, 50 μs – ✓ ✓ Model-based (signal – – ✓
(2013) Iac (FDI in processing in time
power domain
converters)
Solorzano Solórzano and Stand-alone – Varr, Iarr, Parr 1 min PC ✓ – PLA & CDI MATLAB ✓ ✓
Egido (2013)
Meinguet Meinguet et al. – – Iabc phase currents, 1s PC – ✓ Model-based (signal MATLAB ✓ ✓
(2013) Vdc (FDI in processing in frequency
power domain)

171
converters)
Zhao Zhao et al. (2015) Grid tied Gg, Tm Varr, Iarr 1s PC ✓ – ML MATLAB ✓ ✓
Flicker Flicker and Stand-alone/ Gg Rs_ins, Zarr, Ileak, Ifault, – PC ✓ ✓ GFDI Sandia ✓ ✓
Johnson (2013a) Grid tied Iac, Voc, Isc
Hu Hu et al. (2013) Stand-alone/ Gg, Tm Varr, Iarr 1s PC ✓ – HET – ✓ ✓
Grid tied
Chine Chine et al. Grid tied Gg, Tm Varr, Iarr, Parr, Pac 10 min PC ✓ ✓ PLA MATLAB ✓ ✓
(2014)
Ribeiro Ribeiro et al. Stand-alone/ – Varr, Iarr, Parr, Iac, duty 50 μs – ✓ – Digital implementations – – ✓
(2014) Grid tied cycle, Idc, diagnostic (FDI in (based on duty cycle and
signal power Iac derivative sign)
converters)
Nuno Freire et al. – – Vg, Ig, phase currents 0.1 s PC – ✓ Digital implementations MATLAB ✓ ✓
(2014) (FDI in (voltage-based approach)
power
converters)
Campos Campos-Delgado – – Vo, Io – PC – ✓ Model-based (observer MATLAB ✓ ✓
and Espinoza- (FDI in based)
Trejo (2011) power
converters)
Silvestre Silvestre et al. Grid tied Gg, Tm Varr, Iarr 5 min PC ✓ – PLA MATLAB ✓ ✓
(2014)
Alam Alam et al. (2014) Stand-alone/ – ZL 0.1 s – ✓ – SSTDR – – ✓
Grid tied
Luca Bonsignore et al. Stand-alone Gg, Tm Vmp, Imp, Isc, Voc – PC ✓ – ML MATLAB ✓ –
(2014)
(continued on next page)
Solar Energy 158 (2017) 161–185
 Table 4 (continued)

Author Ref. Type of PV Parameters Sampling Monitoring Types of faults considered Fault detection methods Software/ Validation
system period method programming
Meteorological. Electrical. DC side AC side language Simulated Experimental

Cordeiro Cordeiro et al. – – Isc, Phase currents 4 ms – – ✓ Analog implementations – – ✓

S.R. Madeti, S.N. Singh

(2014) (FDI in (based on Z-source

power circuit)
converters)
Yao Yao et al. (2014) Stand-alone/ – Varc, Iarc, IL, 0.1 s PC ✓ – FSA Finite element ✓ ✓
Grid tied analysis
Salehifar Salehifar et al. – – Vo, Io 10 μs PC – ✓ Model-based (Adaptive MATLAB ✓ ✓
(2014) (FDI in model identification)
power
converters)
Nie Nie et al. (2014) – – Gate-driver signal, Isc, – – – ✓ Analog implementations – – ✓
Voc, Vmag (FDI in (based on magnetic
power component)
converters)
Platon Platon et al. (2015 Stand- Gg, Tm – 10 min PC ✓ – ML/ CMM ANN ✓ –
alone & grid
tied
Cho Cho et al. (2015) Stand- alone – Vind 1s PC ✓ – Analog implementations PSIM ✓ ✓
(FDI in
power
converters)
Silvestre Silvestre et al. Grid tied Gg, Tm Varr, Iarr, Parr 5 min PC ✓ ✓ PLA MATLAB ✓ ✓
(2015)
Druant Druant et al. – – Iph, Vbus, Vc, Vdc 20 ms PC – ✓ Model-based (predictive MATLAB ✓ ✓

172
(2015) (FDI in control algorithm)
power
converters)
Chine Chine et al. Stand-alone Gg, Tm Varr, Iarr, Parr, Pac 10 min PC ✓ ✓ PLA MATLAB ✓ ✓
(2015)
Ando Andò et al. (2015) Stand- Gg, Tm Vmod, Imod 10 min PC ✓ – CMM LabVIEW ✓ ✓
alone & grid
tied
Wail Rezgui et al. Stand-alone Gg, Tm Varr, Iarr, Parr – PC ✓ – EM LabVIEW ✓ –
(2015)
Hachana Hachana et al. Stand-alone Gg, Tm Vstr, Istr 30 s PC ✓ – PLA LabVIEW ✓ ✓
(2015) /MATLAB
Bressan Bressan et al. Stand-alone Gg, Tm Vmod, Imod – PC ✓ – EM MATLAB ✓ ✓
(2016)
Chine Chine et al. Stand-alone Gg, Tm Varr, Iarr, Parr – PC ✓ – CMM & ML MATLAB, ✓ ✓
(2016) XSG & ISE
Li Li et al. (2016) – – Vc, Vdc, Io – PC – ✓ Model-based – – ✓
(FDI in
power
converters)
Dhimish Dhimish and Grid tied Gg, Tm Varr, Iarr 1s PC ✓ – CMM LabVIEW ✓ ✓
Holmes (2016)
Liao Liao et al. (2016 Stand-alone Gg, Tm Varr, Iarr – PC ✓ – EM MATLAB ✓ –
Kaplani Kaplani (2016) Stand-alone/ – IR thermography of a – – ✓ – IR/thermal imaging – – ✓
Grid tied PV module
Hachana Hachana et al. Stand-alone Gg, Tm Vstr, Istr 30 s PC ✓ – PLA LabVIEW ✓ ✓
(2016) /MATLAB
Chen Chen et al. (2016) Stand-alone Gg, Tm Varr, Iarr, Voc, Isc 10 min PC ✓ – PLA & CMM MATLAB ✓ ✓
(continued on next page)
Solar Energy 158 (2017) 161–185
 Table 4 (continued)

Author Ref. Type of PV Parameters Sampling Monitoring Types of faults considered Fault detection methods Software/ Validation
system period method programming
Meteorological. Electrical. DC side AC side language Simulated Experimental

Cristina Ventura and Tina Grid tied Gg, Gd, Tm, Ta, Sw Varr, Iarr, Vac, Iac 5 min Web ✓ ✓ CMM PVSyst ✓ ✓
S.R. Madeti, S.N. Singh

(2016)
Mekki Mekki et al. Stand-alone Gg, Tm Vmod, Imod 5 min PC ✓ – ML/ CMM ANN ✓ ✓
(2016)
Silvestre Silvestre et al. Grid tied Gg, Tm Varr, Iarr, Parr 5 min PC ✓ ✓ PLA MATLAB ✓ ✓
(2016)
Chen Chen et al. (2017) Stand-alone Gg, Tm Varr, Iarr 5s PC ✓ – ML/EM MATLAB ✓ ✓
Hazre Hazra et al. Stand-alone Gg, Tm Varr, Iarr 1 min PC ✓ – CMM MATLAB ✓ ✓
(2017)
Guerriero Guerriero et al. Grid tied Gg, Tm Vstr, Istr, Pstr 5 min PC ✓ ✓ CMM – ✓ ✓
(2017)
Grimaccia Grimaccia et al. Stand-alone – Varr, Iarr, Parr – Web ✓ – CDI – – ✓
(2017)
Lin Lin et al. (2017) Grid tied Gg, Tm Varr, Iarr – PC ✓ – ML MATLAB ✓ ✓
Madeti Madeti and Singh Grid tied Gg, Tm Vopt 10 min PC ✓ – CMM – ✓ ✓
(2017b)

Gg: global solar radiation; Gd: diffuse solar radiation; Gt: tilted solar radiation; Tm: module temperature; Ta: ambient temperature; TL: load temperature; Tbat: battery temperature; Sw: wind speed; Varr: array voltage; Iarr: array current; Parr: array
power; Vstr: string voltage; Istr: string current; Pstr: string power; Vmod: module voltage; Imod: module current; Pmod: module power; Pac: output ac power; Vmp: maximum voltage; Imp: maximum current; Pmp: maximum power; Voc: open circuit
voltage; Isc: short circuit current; Il-l: line to line current; Vbat: battery voltage; Ibat: battery current; Vg: grid voltage; VL: load voltage; IL: load current; PC: personal computer; Iarm_conv: converter arm current; Vcell_cap: cell capacitor voltage; Vdc_link:
output dc link voltage; Iph: phase current; Vbus: Bus voltage; Vdc: DC voltage; Io: output current of converter; Vo: output voltage of converter; Tcap: capacitor temperature; ESR: Equivalent series resistance; Xcap: capatitor reactance; RL: load resistance;
Vc1: input dc link capacitor voltage; Iabc: phase currents; Rg: gate driver resistance; Vge: gate emitter voltage; Vge*: gate emitter voltage with attenuation; VT: threshold voltage; Vind: inductor voltage; Vmag: magnetic component voltage; Idc: dc link
current; RDGs: renewable based distributed generators; ϕ: phase angle between voltage and current; ROCOP: rate of change of power; ROCOF: rate of change of frequency; Vopt: optimized voltage sensors reading.

173
Solar Energy 158 (2017) 161–185
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

autocorrelation plots generated by the SSTDR hardware before ground injected current, which provokes an EL effect. The excitation can also
fault and after ground faults at separate locations are presented. obtained by means of a radiation incident over the solar cell, in which
case the light obtained is due to a photoluminescence (PL) effect
5.2. Infrared (IR)/thermal imaging Kasemann et al., 2008. The emission intensity of EL images is propor-
tional to both current density and carrier lifetime, thus it is most
IR/thermal imaging is another popular method for PV fault diag- commonly used for failure analysis. Non-uniform current and poor
nosis. This method is based on the concept of localized heat generation connections are easily detected using this method. The EL images
because of joule heating effect due to shunted cells, poor contacts, short permit the detection of cracks in cells. These images appear dark in the
circuits, etc. In a series connected PV cells, some cells which are gen- images. Furthermore, in cells containing cracks that do not extend all
erating less current as compared to others cells become reverse biased the way through the cell, this kind of test can help to detect potential
and act like resistor and dissipate heat. As a result, temperature gra- problems.
dient is formed which appear as bright spot during thermal imaging.
The thermal imaging can be done in two ways: (i) forward bias imaging 5.6. Lock in thermography (LIT)
(FBI) and (ii) reverse bias imaging (RBI). In case of FBI the module
exposed in the field is connected to a power supply with the module in a Lock in thermography (LIT) is a variation of infrared method. This
forward biased condition. Then the current which is approximately two method is based on finding lateral power loss by injecting current into
times greater than the module short circuit current (specified in the solar cell. This current is not continuous, but rather a pulsed current
data sheet provided by the manufacturer) will flow through the module which causes increase in temperature wherever local shunts are si-
with power supply. It leads to heating of modules. Then after the IR tuated. If the cell has shunt defects, they appear as local temperature
images were recorded using IR camera and perform future image pro- modulations. By using different modulations of the current injection it
cessing to identify the type and location of fault. Defects like hot spot, is easy to detect different kinds of shunt defects. This test can be per-
loose connections and increase in series resistance are identified accu- formed under illumination or in dark conditions. Usually, cells are
rately by using this method. During reverse bias imaging, the same operating under open circuit condition under illumination condition. As
procedure is implemented except the module connected to power the detector is locked with the bias current, small defects can be de-
supply is in reversed biased condition. This process can help to find out tected by using this method. In Breitenstein et al. (2008), to locate a
the ohmic shunts. In a study done by Kaplani (2012), to detect the effect shunts defect, this method is compared with EL and PL methods.
of degradation on the field aged crystalline silicon PV modules from 18
to 22 years used the thermal imaging technique. 6. Fault detection techniques in AC side of PV systems

5.3. Visual inspection Inadvertent islanding and failure of PV inverter are the main tech-
nical issues for the AC side of PV systems. Inverter failure occurs when
Usually, visual inspection is the first step in deciding whether fur- passive components or power switches in the converters are short or
ther tests should conducted on the PV module. This inspection is carried open circuited. High thermal and mechanical stress across the power
out under the 1000 lux light source as per the IEC61215 (International switches can lead to faulty operation of the inverter. Some inverter fault
Electrotechnical Commision, 1987) standards (Standard IE, 1995). In detection techniques proposed in past are based on the power losses
order to avoid the defective images, reflections should be avoided. To analysis (Chine et al., 2014) and machine learning techniques (Yagi
avoid error due to reflected images and also to differentiate the layer et al., 2003). Islanding operation of PV inverter takes place when power
where the defect could appear, the inspection should be done from supply from main grid is interrupted. In this issue, the energy generated
different angles. A detailed visual inspection procedure for operation from the PV array continues to supply a local load or charge controller
and maintenance of PV systems is reported in Haney and Burstein to charge batteries for energy storage. The grid disconnection is usually
(2013). in response to a fault. Many algorithms have been proposed in the past
decade for islanding detection.
5.4. Ultrasonic inspection method
6.1. Fault detection techniques for converter
Fig. 4 shows the methodology used in ultrasonic inspection. This
method is based on the analysis of the ultrasonic vibrations that follow Various fault detection techniques for power electronics system
an excitation. It is used to detect cracks and micro-cracks in cells of PV have been proposed in the past for specific converters. For example, the
module. This method is further classified as (i) pulse-echo method and authors in Rodríguez-Blanco et al. (2011), de Araujo Ribeiro et al.
(ii) transmission method. (2003), Smith et al. (1997), Fang et al. (2015) and Salehifar et al.
In both methods the PV modules is scanned by the moving ultra- (2014), have investigated different fault detection techniques. In other
sonic transducer along with the X-Y indicator assembly. In pulse-echo works (Nie et al., 2014; Pei et al., 2015; Kim et al., 2008; An et al.,
method the PV module is scanned by passing ultrasonic pulses through 2011; Ding et al., 2013; Yazdani et al., 2011), authors have focused on
it and the pulses reflected back from the defects are recorded. The detecting and identifying specific component faults such as capacitor
debonding of cells in a module can be identified by this method. In faults, short and open-circuit switch faults, gate drive faults etc., in dc/
transmission method the attenuated ultrasonic signal is recorded. This dc converters, grid connected dc/ac converters, and in multilevel con-
method helps to locate only X-Y location and size of the defect. For verters. The faults in voltage and current sensors in converter systems
example, in Hund and King (1995) the authors have presented a com- have been discussed in Youssef et al. (2013) and Peuget et al. (1998).
prehensive failure analysis of PV modules after extended field exposure From the literature it is observed that no work has reported a gen-
at various sites. The results have helped to identify the cause of dif- eralized fault detection approach for an arbitrary power converter. A
ferent module degradation mechanisms. majority of existing techniques are specialized for a specific converter
topology and cannot be easily ported from one converter to another.
5.5. Electroluminescence imaging (EL imaging) Methodologies used for fault detection in converters are broadly clas-
sified as: (i) model-based and (ii) model-free approaches. Model-based
Electroluminescence imaging (EL imaging) is obtained from photons approaches are based on residual generation using parameter estima-
emitted by the recombination of excited carriers into a solar cell tion, parity equations, or state observers. These methods utilize analy-
(Kirchartz et al., 2009). The excitation can be achieved by means of an tical knowledge of the system (Jung et al., 2013; Freire et al., 2014;

174
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

Fig. 4. Ultrasonic inspection methodology.

Karimi et al., 2008; Druant et al., 2015; Campos-Delgado and Espinoza- frequency, phase to detect islanding and harmonic distortion. These
Trejo, 2011; Youssef et al., 2013; Lahyani et al., 1998; Ma and Wang, techniques can be further classified into passive, active and hybrid
2005). Signal processing techniques are a subset of these methods. In techniques. The statistic notably shows that active and passive techni-
order to detect abnormalities present in the system, these techniques ques have rapidly increased throughout the years. The details of these
monitor the difference between normal and faulty state of signals. techniques are explained in the following subsections:
Signal processing techniques can either be in the frequency domain
(see, e.g., Shahbazi et al., 2013; Park et al., 2011; Ribeiro et al., 2013; 6.2.1. Central (remote) techniques
Duan et al., 2011) or in the time domain (see, e.g., Karimi et al., 2008 6.2.1.1. System state monitoring. This method is used for determining
and (Lezana et al., 2009; Awadallah and Morcos, 2006; Meinguet et al., system states from the power system network model with a reduced
2013; Peuget et al., 1998; Masrur et al., 2010). number of state measurements. Generally, this method is considered
Model-free approaches rely on learning of particular conditions as a function of the distribution management system (DMS), which is a
corresponding to a particular type of data. Any newly introduced con- complementary of supervisory control and data acquisition (SCADA)
dition can be identified on the basis of previous training (Schapire, systems. By measuring the parameter of the distribution system such as
2008), which can be used for faults detection. Various machine learning voltage and frequency, unintentional islanding of the PV inverter is also
techniques have been employed in the literature such as ANN, fuzzy detected (Balaguer-Alvarez and Ortiz-Rivera, 2010). In Llaria et al.
logic etc. The limitations of these approaches are excessive computa- (2009) and Timbus et al. (2010), the loss of mains has been detected by
tional requirements and large data sets required to train the algorithms. using voltage sensitive devices embedded to PV inverter which are
Finally, the fault detection techniques are further classified into analog connected to SCADA system.
or digital implementations. Analog implementation can detect faults
relatively fast. It will take time around 1–100 μs (Cho et al., 2015;
6.2.1.2. Switch state monitoring. To monitor the status of the reclosers
Cordeiro et al., 2014). However, this method is limited to specific
and circuit breakers (CBs) that could island a distribution system a
converter used and type of fault. Digital implementation provides more
SCADA system can be used. However, this technique requires improved
flexible to user and can be applicable to any kind of converter config-
interaction between the utility and PV inverter which increase the
uration and fault (Ding et al., 2013; Jung et al., 2013; Freire et al., 2014
implementation cost. A transfer trip detection scheme requires all CBs
Sep; Karimi et al., 2008; Druant et al., 2015), etc]. In (Ding et al., 2013;
that island the PV inverter to be monitored and linked directly to the
Druant et al., 2015; Shahbazi et al., 2013), the authors have demon-
inverter control. This scheme is integrated with SCADA to monitor the
strated the speed of fault detection using field-programmable gate array
status of reclosers and CBs (Mahat et al., 2008, 2011; Kunte and Gao,
(hardware implementations) compared to analog implementations.
2008; Llaria et al., 2009; Ezzt et al., 2007). However, the cost of the
system is high because of the system complexity.
6.2. Islanding detection techniques
6.2.1.3. Intertripping. This technique relies on the communication
Various techniques have been developed in the past for islanding between the sensors and the generating units. It detects the opening
detection in grid connected PV inverters. As illustrated in Fig. 5, these of a contact at the points of disconnection and transmits the signal to all
techniques are generally divided into central (remote) and local generation sites. Intertripping provides high reliability and accurate
methods. Remote islanding techniques are based on the communication solution for islanding detection. However, it is not economical
between the utility and the PV inverter. This technique does not have (Chowdhury et al., 2009; Ecconnect, 2001).
any non-detection zone (NDZ), and thus has no impact on PV inverter The above review shows that central control techniques are pre-
power quality. This technique is more effective for multi-inverter sys- ferable because these techniques can avoid NDZ and also these tech-
tems but they require huge investment and complicated communication niques cannot be influenced by the type of generator, number of in-
techniques. Local techniques are based on measurement of system verters, size of system and penetration level. However, a tight
variables or parameters at the grid side, such as current, voltage, connection between generation units and utility grid is needed in

175
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

ANN WT State monitoring

PNN S-transform Switch monitoring

DT Intertripping
Under/over voltage and under/over frequency Voltage unbalance and frequency setpoint
FL Impedance measurement
Voltage phase jump detection
Voltage and real power shift
Slip-mode frequency shift
Harmonic measurements
Voltage fluctuation injection
Active frequency drift
Voltage unbalance

Hybrid SFS and Q-f

Sandia frequency shift

Sandia voltage shift

Fig. 5. Classification of different islanding detection techniques.

applying communication techniques. threshold selection problem and high Q-factor (Kunte and Gao, 2008).
6.2.2.1.4. Voltage unbalance. Generally, the voltage unbalance
varies due to change in topology of the networks and the load despite
6.2.2. Local islanding detection techniques the small changes in the DG load (Mahat et al., 2008, 2011; Jang and
6.2.2.1. Passive techniques Kim, 2004). Therefore, by continuously monitoring the unbalanced
6.2.2.1.1. Under/over voltage and under/over frequency. This is the three-phase output voltage the effective detection of islanding
oldest technique adapted to protect the distribution system. In order to operation is possible. One cycle range average of the voltage
determine various abnormalities the protection relays of this technique unbalanced and voltage unbalanced variation has been defined in
are placed on a distribution feeder. This technique monitors the grid Jang and Kim (2004). It was checked at every ¼ cycle (4.17 ms).
voltage and frequency according to the limits imposed by the IEC61727 From the literature survey, most common attributes in choosing
standards. It is based on the power flow at the point of common passive islanding detection are the simplicity and less cost of im-
coupling (PCC) between the utility and PV inverter. However, the main plementation. For majority of the situation disturbances that occur in
limitation of this method is large NDZ (Zeineldin et al., 2006). the grid passive techniques are effective (Timbus et al., 2010). How-
6.2.2.1.2. Voltage phase jump detection. This method monitors the ever, major drawback of passive techniques is the large NDZ, which
phase difference between the current and the terminal voltage of the fails to detect the islanding condition. The local load affects the de-
inverter for a sudden jump (Hu and Sun, 2009). Then inverter and local termination of the islanding condition. Therefore, those limitations can
load disconnect from the main system when islanding occurs. In order be answered by active techniques.
to detect islanding this method searches for a change in phase angle. It
can be easily implemented because only modified phase locked loop
(PLL) required by the inverters for utility synchronization is needed. 6.2.2.2. Active techniques
However, this method is unsuitable in detecting islanding for all the 6.2.2.2.1. Impedance measurement. This method is also same as the
operating conditions of the load. passive technique, which measures the system impedance changes
6.2.2.1.3. Harmonics measurements. The measurement of line caused by islanding. However, in this method short circuit current
voltage and current harmonics is known as detection of harmonics. and reduced supply voltage are used to calculate the power supply
This method measures the change in total harmonic distortion (THD) at source impedance (Mahat et al., 2008, 2011). Since, this method has no
the PCC. If the THD exceeds the predefined threshold, the PV inverter NDZ, a large number of impedance detection methods have been
should be disconnected from the grid. Since the distortion level changes proposed in literature. In Ropp et al. (2006), an experimental study
as the nonlinear load switches on and off, it does not lead to false has been carried out to verify the impedance detection test. The study
tripping by selecting a trip threshold (Yin, 2005). In Jang and Kim reveals that, this method for islanding detection does have a NDZ.
(2004), the authors proposed two monitoring parameters, namely, THD Therefore, other active techniques can be used to suppress the
of the current and voltage unbalanced to detect the islanding operation limitation of impedance measurement.
of grid. However, this method is difficult to implement because of 6.2.2.2.2. Slip-mode frequency shift (SMS). This method uses

176
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

positive feedback for islanding detection, namely, frequency, amplitude WT can detect the discontinuities and transients in time varying signals
and phase to detect the islanding condition. In this method, a slight for islanding detection. In Pigazo et al. (2007) and Pigazo et al. (2009),
modification of required components can detect the islanding. SMS is the authors proposed a WT technique to obtain the time localization of
also highly effective in islanding prevention compared to other active signals from single-phase PV system. For the islanding mode of
techniques (Zeineldin et al., 2006). It can be implemented easily. operation this method used five decomposition levels. The advantage
However, SMS causes system-level power quality and transient of this system minimizes the computational complexity and reduction
response problems at very high penetration levels and feedback loop of number of sensors required for islanding detection.
gains. This method also introduces a phase shift perturbation which
leads to measurement inaccuracy, noise in response, and quantization 6.2.3.2. S-transform. Under noisy conditions the WT is unable to detect
error. This issue was addressed in Liu et al. (2010), by introducing an the islanding conditions. Therefore, the improved version of WT called
additional phase shift called the improved SMS. The SMS can be S-Transform was introduced (Ray et al., 2010a,b, 2011). S-transform
broadly classified as: (i) Active frequency drift (ii) Sandia frequency produces a time-frequency representation of a time series and has a
shift and (iii) Sandia voltage shift. frequency dependent resolution that simultaneously localizes the real
The literature review shows that the passive islanding techniques and imaginary spectra. It also provides multi-resolution while retaining
have small NDZ . However, the passive techniques have lower reliability the absolute phase of each frequency component. In Ray et al. (2010b,
of than the active techniques. The main limitation in active methods is 2011), to extract the negative sequence voltage during an islanding
the perturbation in the system and the slow detection time. event S-transform was used. The negative sequence voltage and current
processed through the S-transform and spectral energy content has been
6.2.2.3. Hybrid techniques calculated in Samantaray et al. (2010). The drawback of S-transform is
6.2.2.3.1. Voltage unbalance and frequency set point. Menon and the need for more computation time and memory to process the signal
Nehrir (2007), presented a hybrid technique based on the voltage than other techniques.
unbalance and THD and the positive feedback techniques. When both
techniques are applied simultaneously, the limitations of both 6.2.4. Intelligent classifiers
techniques can be cancelled out. In this hybrid technique instead of This technique uses machine learning methods for islanding detec-
THD the voltage unbalance of each distributed generation unit is used tion. The intelligent classifiers are used to learn the relation between
since, the THD is less sensitive to disturbance than the voltage the input and output parameters of the PV inverter and grid and be
unbalance. Therefore, any disturbance applied to generation unit subsequently used to train the models for islanding detection. Some of
could produce a spike in the voltage unbalance. This technique also the machine learning techniques used so far are: ANN (Nie et al., 2014;
differentiates between the islanding condition and load switching Pei et al., 2015; Kim et al., 2008; An et al., 2011; Ding et al., 2013;
efficiently. Yazdani et al., 2011; Youssef et al., 2013; Peuget et al., 1998; Jung
6.2.2.3.2. Technique based on voltage and real power shift. To et al., 2013), probabilistic neural network (PNN) Parzen, 1962; Goh,
overcome the limitations of the active and passive techniques the 2002; Gaing, 2004, and decision trees (DT) Senroy et al., 2006; El-
authors in Mahat et al. (2009), have used an average rate of the real Arroudi et al., 2007; Lidula and Rajapakse, 2010; Pham et al., 2011.
power shift in active technique and the voltage change in passive
technique. This technique can detect the islanding condition with 7. Proposed fault detection technique
multiple generation units which are operating at unity power factor.
In Hung et al. (2003), the proposed model compared the real power Detailed review on various fault detection techniques shown that
changes with the positive feedback technique, where all the generation the modular level fault detection technique provides reliable operation
units work together to inject perturbation in the system. of PV plant. Moreover, well-timed identification of mal-operation at the
6.2.2.3.3. Voltage fluctuation injection. This technique is based on module level is strategically important to guarantee better perfor-
the voltage fluctuation injection, which can be obtained using high mance. Several monitoring systems proposals have already been pre-
impedance load (Chang, 2010). The two stage method, rate of change of sented in the above sections. Traditionally, these systems are based the
voltage/rate of change of current (a passive technique) is used for analysis of data provided by the inverters in conjunction with those
protection scheme, and correlation factor is applied as a backup to provided by a weather station. In some cases, these systems incorporate
attain higher effectiveness. In order to calculate the rate of change of monitoring field lines of panels, data loggers, demand specific software
voltage/rate of change of current and correlation factor, digital signal and personal computer (PC), with the purpose of detecting faults in the
processing is used in this technique. This technique accurately detects PV system. But these methods have certain limitations such as cables for
the islanding and non-islanding disturbances. data transmission, specific skills to operate software, expensive data
6.2.2.3.4. Hybrid Sandia frequency shift (SFS ) and Q-f islanding logger for data collection and the requirement of power supply for PC,
technique. In Vahedi et al. (2010), SFS and Q-f curve is proposed for which increases the cost of the system and limited its spread and use. In
islanding detection. This method improves the SFS and reduces the this section, real time PV module parameters, namely voltage (Vpv ) ,
NDZ. By using analytic formula, the optimum SFS is calculated where current (Ipv ) , and back-panel temperature (Tpv ) monitoring system with
the bacterial foraging algorithm is being used to search the optimal gain fault detection system is proposed. This system uses power line com-
of the SFS to eliminate NDZ. In order to increase the effectiveness of the munication (PLC) technology for data transmission. A new algorithm is
SFS-based islanding technique the Q-f droop curve is added to this developed to detect different types of faults, and to display other es-
technique. sential system information (with a watch window size of 10 min). The
proposed technique is shown to have lower computational require-
6.2.3. Signal processing methods used for identification of island mode ments; therefore, the same microcontroller is used for data commu-
operation nication and fault detection without any external hardware or addi-
6.2.3.1. Wavelet-transform (WT). The wavelet theory is a mathematical tional simulation software.
model for non stationary signals with a set of components in the form of
small waves called wavelets (Polikar, 1999). Using wavelet is 7.1. Proposed monitoring system architecture
advantageous because the wavelet does not assume the stationary or
periodicity of signal. Since wavelet has short windows at high Fig. 6 illustrates the architecture of proposed monitoring system. It
frequencies and high windows in short frequencies, it is capable of consists of low-cost microcontroller-based smart monitoring system
comprehending frequency and time information simultaneously. Thus, (SMS) installed in PV array consisting of PV modules connected in

177
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

TFT-LCD

SD CARD
Bypass diode

I1 V1
Slave Master
Blocking diode
PLC module PLC module
Bypass diode

I2 V2 Iinv Vinv

PV ARRAY INVERTER GRID

Fig. 6. Block scheme of the Monitoring system architecture.

series, data transmission path, and the designed tool for system super- (i) Expected PV array efficiency: In order to completely avoid the
vision. This transmission path uses PLC technology that enables com- simulation tool in fault detection procedure the values of voltage
munication over already existing dc power lines (red dotted lines), and and current of PV module at MPP is calculated analytically. For an
requires no additional installation cost. arbitrary value of G and T , the maximum current and voltage of a
PV array of arbitrary series-parallel (Ns × Np ) connection of PV
7.2. Proposed fault detection technique modules can be given as

I dI
The fault detection approach suggested in this paper is mainly fo- Im = Np ⎛ mmr G + ⎛ scm ⎞ (Tcell−Tr )⎞
⎜ ⎟

⎝ 1000 ⎝ dT ⎠ ⎠ (5)
cused on the development of paradigms for the assessment of accidental
causes leading to energy losses, such as partial shading, open or short I −I Vocm
circuit fault in any module, bypass diode fault and faulty inverter. This Vm = Ns ⎜⎛Nsc VT ln ⎜⎛1 + scm mm ⎛e Ns VT −1⎞ ⎟⎞−Imm Rsm⎞⎟
⎜ ⎟

⎝ ⎝ Iscm ⎝ ⎠⎠ ⎠ (6)
technique is based on the comparison between the measured and model
prediction results of the PV array and inverter efficiency to detect the The measured and calculated efficiency of PV array is defined as
energy losses. The predictive model used in this approach was selected follows
such that its development does not require knowledge of advanced Vmeas × Imeas
modeling methods, artificial intelligence techniques, detailed compre- ηpv meas =
G × Ac (7)
hension of the PV process and electrical circuits necessary to carry out a
circuit-based simulation, or familiarity with commercial energy simu- Vpv cal × Ipv cal
ηpv cal =
lation software packages. For proper operation of the fault detection G × Ac (8)
system, specific threshold value for PV array and inverter efficiency (ηpv
and ηinv ) is essential. It avoids chances of false or missed fault detection. where Vmeas , Imeas and G are the measured voltage, current and irra-
Hence, threshold for these efficiencies are established in this work. If diance, respectively, while Ac is the area of the PV collector.
the values are lower than the predefined threshold, then it is considered
as PV system is normally operating, otherwise it is considered that an (ii) Expected inverter efficiency
abnormality is present in the system. The statistical analysis of ηpv and
ηinv have been made during the absence of fault in PV system and es- The efficiency of an inverter converting the DC current and voltage
tablished the reference thresholds. Then, standard deviation (σ ) is de- into the AC current and voltage has been modelled with the standard
rived for both ηpv and ηinv . It is identified that the PV system works in approach given by the following equation. The model describes the
fault free operation, when the ηpv and ηinv values are in between the efficiency as a function of the input power (Pin ) So et al., 2007.
following threshold: ηinv = y0 + c1 (1−exp (−c2 × Pin )) + c3 (1−exp (−c4 × Pin )) (9)
Δηpv ref −σ ⩽ Δηpv ⩽ Δηpv ref + σ (3) where y0 , c1, c2, c3, and c4 are five parameters that can be determined
using a non-linear regression algorithm, such as the least square algo-
Δηinv ref −σ ⩽ Δηinv ⩽ Δηinv ref + σ (4) rithm. Fig. 7 shows a flow chart of the proposed modular level PV di-
agnosis system.

7.3. Predicted PV array and inverter efficiency 7.4. Verification of proposed fault detection technique

The prediction of PV array and inverter efficiency is respectively The grid connected photovoltaic (GCPV) system described above
explained in points (i) and (ii) was monitored on a clear day, 9th August 2017 in fault free operation.

178
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

Fig. 7. Flowchart of the proposed PV fault diagnosis system.

The evaluation of efficiency of PV array calculated ηpv cal was done using approximately in 1.57 years (∼19 months) of PV plant operation.
Eq. (7). It is observed that both measured PV array efficiency is very
close to calculated one, as is clearly shown in Fig. 8(a). The normality/ 8. Discussion of different fault detection techniques
abnormality of the system are given by the state of diagnosis signal. In
this case, it is in low “0” state, indicating that there is no-fault present in As mentioned in previous sections, there are many techniques for
the GCPV system. One can conclude that the system is operating nor- detection of faults in the PV systems. The important findings of these
mally and further information about the performance of the system techniques for DC and AC side of PV system are summarized separately
indicated in the designed tool is coherent. Five faults were created on in the Sections 7.1 and 7.2 respectively. Under each subsection, the
day 2 (10th, August 2017) that are collectively shown in Fig. 8(b). In discussions are given in a point wise manner.
order to test the effectiveness of the proposed technique short circuit
and partial shading faults are considered in this study. To create these 8.1. Fault detection in DC side of PV system
two faults, the terminal of one PV module was shorted and one module
was partially covered with transparent gelatin paper, respectively. i. In climatic data independent (CDI) techniques such as TDR, HET,
Fig. 8(c) and (d) shows the details about the fault identified in the Thin ECM and thermal cameras are popular methods for fault diagnosis.
Film Transistor (TFT) display. When compared to proposed techniques However, a gradual change in the thermal image of a PV module
in the literature, proposed technique exactly located the location and poses a technical challenge, and also high system cost. Other
type of fault present in the system. Though, the proposition in this work techniques HET, TDR and ECM can operate in offline mode only. In
will slightly increase the initial cost. However, it will be easily com- practice, online fault detection techniques are highly desired.
pensated by long term healthy operation of the system. ii. To improve off-line fault detection techniques, an automatic su-
pervision and fault detection techniques such as EM, CMM, PLA,
ML are proposed. The important findings of these techniques are
7.5. Cost analysis of the monitoring system summarized in points (a)–(d).
a. In PLA, the fault detection is based on the rate of energy losses
It has been demonstrated that the use of PLC communication tech- in PV system. This technique requires surrounding environ-
nology on dc lines is possible avoiding additional installation costs. The mental information and it cannot identify the faulty module.
use of standard industrial communication interfaces (RS232, RS485) b. A comparison of expected values of PV parameters with real
require additional wiring, with an additional cost of INR . 55,000 or time measurements is then developed. Usually, these techniques
radio systems based on wireless sensors network (+INR. 30,000). (EM, CMM) can establish fixed theoretical thresholds, which the
In a 100-kW PV which can potentially invoice INR . 11,62,500/year, change in systems output power does not exceed, otherwise the
an accumulated continuous 10% undetected losses would represent the system is considered in faulty operation. Since, the output
loss of revenue of INR. 1,16,500/year. The estimation cost of SMS is power is highly dependent on climatic conditions it is difficult
approximately INR. 421.95 (which is a 3.5% of 230Wp PV panel), a 435 to set fixed threshold values for protective devices. These
PV plant monitoring system deployed would represent an investment of methods can locate open and short circuit faulty but still cannot
roughly INR. 1,83,550. The rate of the investment is done

179
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

Fig. 8. Observations in the designed tool: (a) PV array measured and calculated efficiency results on 9th August 2017 (b) Five faults observed in the designed tool 10th August 2017 (c)
Short circuit fault in PV module is detected (10th August 2017) (d) Partial shading fault in module is detected (10th August 2017).

identify the faulty module from the array. measurements during inverter operation for ungrounded systems
c. Thus, a learning technique is required to set their own threshold only. Conversely, this technique can be influenced by the me-
values by training the system based on observed data. The teorological variations. For grounded systems, while taking mea-
training data will vary depending on the electrical properties of surements the inverter and GFDI fuse need to be disconnect.
PV panel and climatic conditions of the installed PV system. vi. FSA of voltage and current waveforms technique is widely re-
d. Currently, both offline and online fault detection techniques are commended for arc fault detection. The accuracy of this technique
available. However, these techniques require surrounding en- is very high. The decision taken by this technique is based on the
vironmental information, and data loggers to collect meteor- broadband spectrum instead of one/few frequencies. However,
ological and electrical parameters, demand special software and this method requires Fourier transformation of signals. Noise from
require power supply or a personal computer to be connected power converter may lead to false tripping. Since this technique is
all the time, which increases the cost and requiring additional based on frequency analysis, it cannot differentiate between the
software operating skills. parallel and series arc faults.
iii. During fault condition, system grounding (GFDI) provides a cir- vii. ERV for arc fault detection is easy to implement. It has less com-
culating path for the ground current. By using this method, no putational cost compared to Discrete Wavelet Transformation and
electric shock could be felt due to parasitic capacitors. This method Fast Fourier Transform. Conversely, the threshold for variance to
is less expensive compared to IMD and RCD techniques. However, alarm arc need to be precise. And also, it needs efficient FIR filter
it can be used in grounded PV system only. This technique needs estimator for calculating variance.
external sensor to monitor whether the fuse is cleared. viii. SSTDR technique does not require voltage and current measure-
iv. RCD technique can be used for both grounded and ungrounded ment of the PV array for arc fault detection. This method can be
systems. RCD can be implemented easily. This technique can take tested during the absence of sunlight. By detecting the change in
the measurements during inverter operation. However, external resistance of PV strings, the future arc faults can be predicted.
noise may result in false tripping. This technique can be useful in However, this technique requires high sampling speed of mea-
corrosion and electric shock hazards. It cannot differentiate be- surements. Noise due to inverters may leads to false tripping. It
tween the ground and line-to-line faults. requires a baseline of healthy PV array for comparison.
v. IMD technique for ground fault detection is more reliable. It can be ix. Non destructive IR/thermal imaging is capable to detect Hot spot
tested during the absence of sunlight. This technique can take generation, and increase in series resistance. Ultrasonic imaging is

180
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

able to locate air voids, delamination and debonding which is not easy to implement but, their reaction times are variable and
visible by naked eye. LIT, computed tomography using X-rays are unpredictable. These methods have large NDZ. For constant
suitable for studying reliability and failure analysis issues like μ- current controlled inverter, it is compared with P-V and P-Q
cracks in Si cells of PV module. Finally, EL imaging is helpful in models.
distinguishing the decrease in shunt resistance and increase in 2. Voltage phase jump detection technique is controlled by using
series resistance. These two defects can leads to hotter areas in IR PLL. It is difficult to implement and hard to chose threshold
imaging which is difficult to detect. value. This method fails to detect islanding when power
generation meets the local load.
From the above discussion it is observed that still there is a scope to 3. Harmonic measurement is easy to implement but hard to
develop an effective yet low cost module level PV monitoring system choose threshold value. This method failed to detect islanding
along with efficient on-line fault detection technique. To improve the in case of high quality load or low distortion of current and
existing monitoring systems with fault detection technique, some future voltage output of inverter.
improvements are suggested here as follows: 4. Voltage unbalance techniques are not applicable to single
phase systems. This technique can be improved by combining
• Optimize the sensor locations that can effectively locate faulty PV two or more methods. For example, unbalanced voltage and
strings and PV modules. THD of voltage and current magnitude variation.
• Sensors failure detection may be added to the developed system to b. Active islanding techniques: These techniques reduce the
improve the robustness of the system. size of NDZ and decrease error detection ratio. However, these
• Develop a fault monitoring system which should have autonomous methods are inevitable to maintain good power quality and are
operation with low energy consumption and is independent of a usually applied to detect grid islanding. The following points
computer or has access to a power grid, which decreases the cost (1)–(2) summarizes the existing major issues when using the
and complexity of the system. active islanding techniques.
• Faulty notification to user by using messaging system may become 1. Impedance measurement islanding technique is easy to im-
an added value to the system. plement and fast in response. However, it has large NDZ for
high reactive loads.
8.2. Fault detection in AC side of PV system 2. SMS is hard to implement compared to Impedance measure-
ment method. Its response speed is slow. This method is in-
i. Model based techniques use analytical knowledge of system. These efficient under creation loads such as RLC resonant load etc.
techniques take long time (10–20 ms) to identify faults depending However, this method also has large NDZ for high reactive
on the complexity of the computational algorithm and also latency loads. The other subclassification methods are: (i) active fre-
of the computing platform. quency drift method is easy to implement and it has medium
ii. Model-free techniques rely on machine learning techniques. These response time. It has large NDZ for high reactive loads (ii)
methods can set their own threshold values by training the system Sandia frequency shift method is difficult to implement and
based on observed data. However, these techniques require large fast in response. It has problem in system stability and power
dataset to train the algorithm and excessive computational re- quality. NDZ exist for high reactive loads but less compared to
quirements. others. (iii) Sandia voltage shift is medium to implement and
iii. Analog implementations can identify faults relatively fast fast in response. This method increases the harmonic distor-
(1–100 μs). However, these techniques are both converter and fault tion. It has very small NDZ.
specific. Digital implementation can provide more flexible in terms c. Hybrid techniques: These techniques are effective to be ap-
of converter topology and type of fault. plied in complex systems. It can improve performance indices
iv. Remote islanding techniques have no NDZ and have no impact on with the combination of different methods. The summary of the
system transient response and power quality. These techniques can passive islanding techniques is given in points (1)–(3).
eliminate the error detection. However, it requires large investment 1. Voltage unbalance and frequency set point hybrid technique
cost because it is located in remote areas. These techniques are has small NDZ and error detection rate is very low. It has small
applied in high-density distributed generation systems. The sum- impact on power quality.
mary of the remote islanding techniques are given in following 2. Technique based on voltage and real power shift has small
points (a)–(c). NDZ and it has low error detection rate.
a. SCADA system for islanding detection has no NDZ and no impact 3. Hybrid Sandia frequency shift (SFS) and Q-f islanding tech-
on power quality. If the system is less processing load then the nique has very small NDZ. It has no error detection rate and
islanding detection speed is high. Otherwise, if the system is has reduced negative impact on power quality. This technique
busy then detection time is very slow. It is highly effective in is effective for multi-inverter cases.
multiple inverter cases. However, the implementation cost is vi. The implementation of signal processing in islanding techniques
high and it is difficult to maintain. extracts the hidden features of the measured signals. These ex-
b. Transfer trip technique is a simple concept and it avoids NDZ. tracted features can be input to the machine learning classifier.
However, direct transfer trip can avoid islanding These methods can easily analyze the signals. The points (a)–(c)
c. Wired cable or non-wired cable for islanding detection approach summarizes the signal processing techniques in islanding detection.
is east to implement. But, the investment cost is high. Moreover, a. In continuous wavelet signal processing method, voltage, cur-
this techniques requires possible media to transfer the signals. rent, and power signals are measured for islanding detection.
v. Local islanding techniques have no impact of power quality. The key indicator for this method is modules local maxima. Its
a. Passive islanding techniques: These techniques have no takes 0.6 s (rate of change of frequency) approximately for is-
impact on power quality and detection speed is fast. However, landing detection.
the error detection rate and NDZ is high compared to active b. In discrete wavelet signal processing method, voltage, current,
methods. These techniques are generally applied to single gen- and power signals are measured along with terminal voltage and
eration system with a certain power imbalance. Major issues of current at PCC of generation units, change in energy and stan-
the passive islanding techniques are given in points (1)–(4). dard deviation, and spectral change of PCC voltage are also
1. Under/over voltage and under/over frequency techniques are measured. The key indicators for this method are d1 coefficients,

181
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

third and fifth decomposition level, NROCOPI, energy values of Aljankawey AS, Morsi WG, Chang L, Diduch CP. Passive method-based islanding detec-
tion of renewable-based distributed generation: the issues. In: Electric Power and
second level wavelet coefficients. It takes less than 2 s (2.5 Energy Conference (EPEC), 2010 IEEE 2010:1–8.
power frequency cycles) or less than 0.2 s (one third of cycles for Amaral, A.M., Cardoso, A.J., 2012. On-line fault detection of aluminium electrolytic ca-
60 Hz) depending upon the type of parameters and key in- pacitors, in step-down DC-DC converters, using input current and output voltage
ripple. IET Power Electron. 5 (3), 315–322.
dicators used for islanding detection. An, Q.T., Sun, L.Z., Zhao, K., Sun, L., 2011. Switching function model-based fast-diag-
c. In S-transform technique, voltage signal is only measured for nostic method of open-switch faults in inverters without sensors. IEEE Trans. Power
islanding detection. The key indicators are coherency, energy Electron. 26 (1), 119–126.
Andò, B., Baglio, S., Pistorio, A., Tina, G.M., Ventura, C., 2015. Sentinella: Smart mon-
and standard deviation indices. itoring of photovoltaic systems at panel level. IEEE Trans. Instrum. Meas. 64 (8),
vii. In intelligent classifier, the information generated by the type of 2188–2199.
signal processing can be used to identify the island mode of op- Awadallah, M.A., Morcos, M.M., 2006. Automatic diagnosis and location of open-switch
fault in brushless DC motor drives using wavelets and neuro-fuzzy systems. IEEE
eration. By using wavelet and discrete wavelet transform signal
Trans. Energy Convers. 21 (1), 104–111.
processing methods in ANN classifier, the recognition rate will be Bahrani, B., Karimi, H., Iravani, R., 2011. Non detection zone assessment of an active
95.0 and 90.0, respectively. If discrete wavelet transform signal islanding detection method and its experimental evaluation. IEEE Trans. Power
processing is used in DT classification method then the recognition Delivery 26 (2), 517–525.
Balaguer-Alvarez, I.J., Ortiz-Rivera, E.I., 2010. Survey of distributed generation islanding
rate will be 94.0. detection methods. IEEE Latin Am. Trans. 8 (5), 565–570.
Ball, G., Brooks, B., Johnson, J., Flicker, J., Rosenthal, A., Wiles, J., Sherwood, L., Albers,
9. Conclusion M., Zgonena, T., 2013. Inverter ground-fault detection ‘blind spot’ and mitigation
methods. Solar Amer. Board Codes Stand. Rep. (June).
Benghanem, Mohamed, Maafi, Abdelbaki, 1998. Data acquisition system for photovoltaic
In this paper an attempt has been made to discuss various faults and systems performance monitoring. IEEE Trans. Instrum. Meas. 47 (1), 30–33.
fault detection techniques present in DC and AC side of the PV systems. Bonsignore, L., Davarifar, M., Rabhi, A., Tina, G.M., Elhajjaji, A., 2014. Neuro-fuzzy fault
detection method for photovoltaic systems. Energy Procedia 62, 431–441.
Generally, for fault detection in DC side of PV system both offline and Bower, W., Wiles, J., 2000. Investigation of ground-fault protection devices for photo-
online based models are used. Thermal cameras, earth capacitance voltaic power system applications. In: Photovoltaic Specialists Conference, IEEE, pp.
measurements (ECM), and time-domain reflectometry (TDR) are three 1378–1383.
Braun, H., Buddha, S.T., Krishnan, V., Spanias, A., Tepedelenlioglu, C., Yeider, T.,
popular off-line methods for fault diagnosis. It is observed that ECM
Takehara, T., 2012. Signal processing for fault detection in photovoltaic arrays. In:
technique can be used to detect disconnection of PV module, whereas Acoustics, Speech and Signal Processing (ICASSP), IEEE International Conference, pp.
TDR is used to detect the disconnection of PV string as well as the 1681–1684.
Breitenstein, O., Bauer, J., Trupke, T., Bardos, R.A., 2008. On the detection of shunts in
impedance change due to degradation. Another technique, thermal
silicon solar cells by photo-and electroluminescence imaging. Prog. Photovoltaics
camera is used to identify the degradation/failure site in a field exposed Res. Appl. 16 (4), 325–330.
module. In order to take measurements while system is in operation Bressan, M., El Basri, Y., Galeano, A.G., Alonso, C., 2016. A shadow fault detection
online fault detection techniques are developed. Various online fault method based on the standard error analysis of IV curves. Renewable Energy 99,
1181–1190.
detection techniques have been discussed in this paper. To prevent fire Brooks, B. 2012.“The ground-fault protection blind spot: Safety concern for larger PV
in grounded and ungrounded PV arrays, different commercial fault systems in the U.S.”, Solar Amer. Board Codes Stand. Rep., Jan.
detection and mitigation techniques are studied in detail in this paper. Campos-Delgado, D.U., Espinoza-Trejo, D.R., 2011. An observer-based diagnosis scheme
for single and simultaneous open-switch faults in induction motor drives. IEEE Trans.
However, ground faults may remain undetected with conventional de- Industr. Electron. 58 (2), 671–679.
tection techniques due to some limitations. It causes severe damage to Chan, F., Calleja, H., 2006. Reliability: A new approach in design of inverters for PV
the surrounding environment and PV array. In case of line-to-line faults systems. In: International Power Electronics Congress, IEEE, pp. 1–6.
Chang, W.Y., 2010. A hybrid islanding detection method for distributed synchronous
similar situations may arise. An overview of different fault detection generators. In: Power Electronics Conference (IPEC), pp. 1326–1330.
and mitigation techniques for arc faults along with recommendations Chao, K.H., Ho, S.H., Wang, M.H., 2008. Modeling and fault diagnosis of a photovoltaic
for future improvements are also presented in this paper. system. Electr. Power Syst. Res. 78 (1), 97–105.
Chao, K.H., Chen, C.T., Wang, M.H., Wu, C.F., 2010. A novel fault diagnosis method
The fault detection for power electronics and various islanding de- based-on modified neural networks for photovoltaic systems. Adv. Swarm Intell.
tection techniques explored in the literature have been presented. The 531–539.
necessary features of islanding detection techniques and the potential of Chen, Y.H., Liang, R., Tian, Y., Wang, F., 2016. A novel fault diagnosis method of PV
based-on power loss and IV characteristics. IOP Conf. Series: Earth Environ. Sci. 40
local and remote techniques used in the system have been investigated.
(1), 012022.
Various intelligent system applications in islanding detection have been Chen, Z., Wu, L., Cheng, S., Lin, P., Wu, Y., Lin, W., 2017. Intelligent fault diagnosis of
discussed. Finally, a new fault detection and diagnosis technique is photovoltaic arrays based on optimized kernel extreme learning machine and IV
proposed. The type and exact location of fault was properly detected characteristics. Appl. Energy.
Cheng, Z., Zhong, D., Li, B., Liu, Y., 2011. Research on fault detection of PV array based
using this technique. Hence the proposed technique can be used in fault on data fusion and fuzzy mathematics. In: Power and Energy Engineering Conference
detection applications. Additionally, this review will help the re- (APPEEC), pp. 1–4.
searches for developing an economic and effective fault detection Chine, W., Mellit, A., Pavan, A.M., Kalogirou, S.A., 2014. Fault detection method for grid-
connected photovoltaic plants. Renewable Energy 66, 99–110.
technique viable for even small and medium scale PV plants. Chine, W., Mellit, A., Pavan, A.M., Lughi, V., 2015. Fault diagnosis in photovoltaic arrays.
In: Clean Electrical Power (ICCEP), International Conference on 2015, pp. 67–72.
Acknowledgement Chine, W., Mellit, A., Lughi, V., Malek, A., Sulligoi, G., Pavan, A.M., 2016. A novel fault
diagnosis technique for photovoltaic systems based on artificial neural networks.
Renewable Energy 90, 501–512.
One of the authors (SRK) acknowledges the financial support of Cho, H.K., Kwak, S.S., Lee, S.H., 2015. Fault diagnosis algorithm based on switching
Ministry of New and Renewable Energy (MNRE), Government of India, function for boost converters. Int. J. Electron. 102 (7), 1229–1243.
Chouder, A., Silvestre, S., 2010. Automatic supervision and fault detection of PV systems
New Delhi, India in the form of scholarship. The authors also like to based on power losses analysis. Energy Convers. Manage. 51 (10), 1929–1937.
acknowledge Assoc. Prof. Jan Kleissl whose comments greatly im- Chowdhury, S.P., Chowdhury, S., Crossley, P.A., 2009. Islanding protection of active
proved this work. distribution networks with renewable distributed generators: A comprehensive
survey. Electr. Power Syst. Res. 79 (6), 984–992.
Ciampi, A., Appice, A., Malerba, D., Muolo, A., 2012. An intelligent system for real time
References fault detection in PV plants. In: Sustainability in Energy and Buildings, pp. 235–244.
Cordeiro, A., Palma, J.C., Maia, J., Resende, M.J., 2014. Detection and diagnosis solutions
Alam, M.K., Khan, F.H., Johnson, J., Flicker, J., 2014. PV arc-fault detection using spread for fault-tolerant VSI. J. Power Electron. 14 (6), 1272–1280.
spectrum time domain reflectometry (SSTDR). In: Energy Conversion Congress and Davarifar, M., Rabhi, A., El Hajjaji, A., 2013. Comprehensive modulation and classifica-
Exposition (ECCE), IEEE conference, pp. 3294–3300. tion of faults and analysis their effect in DC side of photovoltaic system. Energy
Alam, M.K., Khan, F., Johnson, J., Flicker, J., 2015. A comprehensive review of cata- Power Eng. 5 (04), 230.
strophic faults in PV arrays: types, detection, and mitigation techniques. IEEE J. de Araujo Ribeiro, R.L., Jacobina, C.B., da Silva, E.C., Lima, A.M., 2003. Fault detection of
Photovolt. 5 (3), 982–997. open-switch damage in voltage-fed PWM motor drive systems. IEEE Trans. Power

182
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

Electron. 18 (2), 587–593. of DC-AC inverter interfaced DG systems. In: Universities Power Engineering
Dhimish, M., Holmes, V., 2016. Fault detection algorithm for grid-connected photovoltaic Conference (UPEC), pp. 1–5.
plants. Sol. Energy 137, 236–245. Hazra, A., Das, S., Basu, M., 2017. An efficient fault diagnosis method for PV systems
Ding, X., Poon, J., Čelanović, I., Domínguez-García, A.D., 2013. Fault detection and following string current. J. Clean. Prod. 154, 220–232.
isolation filters for three-phase AC-DC power electronics systems. IEEE Trans. Circuits Heidari, N., Gwamuri, J., Townsend, T., Pearce, J.M., 2015. Impact of snow and ground
Syst. I Regul. Pap. 60 (4), 1038–1051. interference on photovoltaic electric system performance. IEEE J. Photovolt. 5 (6),
Dini, D.A., Brazis, P.W., Yen, K.H., 2011. Development of arc-fault circuit-interrupter 1680–1685.
requirements for photovoltaic systems. In: Photovoltaic Specialists Conference Hernandez, J.C., Vidal, P.G., 2009. Guidelines for protection against electric shock in PV
(PVSC), 2011 37th IEEE, pp. 001790–001794. generators. IEEE Trans. Energy Convers. 24 (1), 274–282.
Drews, A., De Keizer, A.C., Beyer, H.G., Lorenz, E., Betcke, J., Van Sark, W.G., Hernández, J.C., Vidal, P.G., Medina, A., 2010. Characterization of the insulation and
Heydenreich, W., Wiemken, E., Stettler, S., Toggweiler, P., Bofinger, S., 2007. leakage currents of PV generators: Relevance for human safety. Renewable Energy 35
Monitoring and remote failure detection of grid-connected PV systems based on sa- (3), 593–601.
tellite observations. Sol. Energy 81 (4), 548–564. Houssein, A., Heraud, N., Souleiman, I., Pellet, G.,2010. Monitoring and fault diagnosis of
Drif, M., Mellit, A., Aguilera, J., Pérez, P.J., 2012. A comprehensive method for estimating photovoltaic panels. In: Energy Conference and Exhibition (EnergyCon), 2010 IEEE
energy losses due to shading of GC-BIPV systems using monitoring data. Sol. Energy International 2010, pp. 389–394.
86 (9), 2397–2404. Hu, W., Sun, Y.L.. 2009. A compound scheme of islanding detection according to inverter.
Druant, J., Vyncke, T., De Belie, F., Sergeant, P., Melkebeek, J., 2015. Adding inverter In: Power and Energy Engineering Conference, Asia-Pacific, pp. 1–4.
fault detection to model-based predictive control for flying-capacitor inverters. IEEE Hu, Y., Gao, B., Song, X., Tian, G.Y., Li, K., He, X., 2013. Photovoltaic fault detection using
Trans. Industr. Electron. 62 (4), 2054–2063. a parameter based model. Sol. Energy 96, 96–102.
Duan, P., Xie, K.G., Zhang, L., Rong, X., 2011. Open-switch fault diagnosis and system Hu, Y., Zhang, J., Cao, W., Wu, J., Tian, G.Y., Finney, S.J., Kirtley, J.L., 2015. Online two-
reconfiguration of doubly fed wind power converter used in a microgrid. IEEE Trans. section PV array fault diagnosis with optimized voltage sensor locations. IEEE Trans.
Power Electron. 26 (3), 816–821. Industr. Electron. 62 (11), 7237–7246.
Ducange, P., Fazzolari, M., Lazzerini, B., Marcelloni, F., 2011. An intelligent system for Hund, T.D., King, D.L., 1995. Analysis techniques used on field degraded photovoltaic
detecting faults in photovoltaic fields. In: Intelligent Systems Design and Applications modules. NASA STI/Recon Technical Report N, p.96.
(ISDA), 11th International Conference on, pp. 1341–1346. Hund, T.D., King, D.L., 1995. Analysis techniques used on field degraded photovoltaic
Ecconnect, 2001. Assessment of islanded operation of distribution networks and measures modules. NREL photovoltaics performance and reliability workshop.
for protection. In: Polymer contents, pp. 830–877. Hung, G.K., Chang, C.C., Chen, C.L., 2003. Automatic phase-shift method for islanding
El-Arroudi, K., Joos, G., Kamwa, I., McGillis, D.T., 2007. Intelligent-based approach to detection of grid-connected photovoltaic inverters. IEEE Trans. Energy Convers. 18
islanding detection in distributed generation. IEEE Trans. Power Delivery 22 (2), (1), 169–173.
828–835. Instruments, T., 2012. AN-2154 RD-195 DC Arc Detection Evaluation Board. TI, Dallas.
ElNozahy, M.S., El-Saadany, E.F., Salama, M.M., 2011. A robust wavelet-ANN based International Electrotechnical Commission, 1998. IEC 61724, Photovoltaic System
technique for islanding detection. In: Power and Energy Society General Meeting, Performance Monitoring e Guidelines for Measurement, Data Exchange and Analysis,
2011 IEEE, pp. 1–8. first ed. International Electrotechnical Commission, Geneva, IEC.
Ezzt, M., Marei, M.I., Abdel-Rahman, M., Mansour, M.M., 2007. A hybrid strategy for Jang, S.I., Kim, K.H., 2004. An islanding detection method for distributed generations
distributed generators islanding detection. In: Power Engineering Society Conference using voltage unbalance and total harmonic distortion of current. IEEE Trans. Power
and Exposition in Africa, IEEE Power Africa, pp. 1–7. Delivery 19 (2), 745–752.
Ezzt, M., Marei, M.I., Abdel-Rahman, M., Mansour, M.M., 2007. A hybrid strategy for Jiang, L.L., Maskell, D.L., 2015. Automatic fault detection and diagnosis for photovoltaic
distributed generators islanding detection. In: Power Engineering Society Conference systems using combined artificial neural network and analytical based methods. In:
and Exposition in Africa, PowerAfrica'07, pp. 1–7. Neural Networks (IJCNN), IEEE International Joint Conference, vol. 12, pp. 1–8.
Fang, J., Li, W., Li, H., Xu, X., 2015. Online inverter fault diagnosis of buck-converter Johnson, J., Kang, J., 2012. Arc-fault detector algorithm evaluation method utilizing
BLDC motor combinations. IEEE Trans. Power Electron. 30 (5), 2674–2688. prerecorded arcing signatures. In: 38th IEEE Photovoltaic Specialists Conference
Fayyad, Y., Osman, A., 2010. Neuro-wavelet based islanding detection technique. In: (PVSC), pp. 001378–001382.
Electric Power and Energy Conference (EPEC), 2010 IEEE, pp. 1–6. Johnson, J., Pahl, B., Luebke, C., Pier, T., Miller, T., Strauch, J., Kuszmaul, S., Bower, W.,
Firth, Steven K., Lomas, Kevin J., Rees, Simon J., 2010. A simple model of PV system 2011. Photovoltaic DC arc fault detector testing at Sandia National Laboratories. In:
performance and its use in fault detection. Sol. Energy 84 (4), 624–635. Photovoltaic Specialists Conference (PVSC), 2011 37th IEEE, pp. 003614–003619.
Flicker, J., Johnson, J., 2013. Analysis of fuses for Blind Spot ground fault detection in Johnson, J., Kuszmaul, S., Bower, W., Schoenwald, D., 2011. Using PV module and line
photovoltaic power systems. Solar Am. Board Codes Stand Rep., (June). frequency response data to create robust arc fault detectors. In: Proceedings of the
Flicker, J., Johnson, J., 2013. Photovoltaic ground fault and blind spot electrical simu- 26th European Photovoltaic Solar Energy Conference and Exhibition, pp. 05–09.
lations. Sandia Nat. Lab., Albuquerque, NM, USA, Tech. Rep. SAND2013-3459 Jung, S.M., Park, J.S., Kim, H.W., Cho, K.Y., Youn, M.J., 2013. An MRAS-based diagnosis
(June). of open-circuit fault in PWM voltage-source inverters for PM synchronous motor
Flicker, J, Johnson, J., 2013. Electrical simulations of series and parallel PV arc-faults. In: drive systems. IEEE Trans. Power Electron. 28 (5), 2514–2526.
Photovoltaic Specialists Conference (PVSC), 2013 IEEE 39th 2013, pp. 3165–3172. Kaplani, E., 2012. Detection of degradation effects in field-aged c-Si solar cells through IR
Freire, N.M., Estima, J.O., Cardoso, A.J., 2014 September. A voltage-based approach thermography and digital image processing. Int. J. Photoenergy.
without extra hardware for open-circuit fault diagnosis in closed-loop PWM AC re- Kaplani, E., 2016. Degradation in field-aged crystalline silicon photovoltaic modules and
generative drives. IEEE Trans. Industr. Electron. 61 (9), 4960–4970. diagnosis using electroluminescence. Imaging.
Gaing, Z.L., 2004. Wavelet-based neural network for power disturbance recognition and Kaplanis, S., Kaplani, E., 2011. Energy performance and degradation over 20years per-
classification. IEEE Trans. Power Delivery 19 (4), 1560–1568. formance of BP c-Si PV modules. Simul. Model. Pract. Theory 19 (4), 1201–1211.
Goh, A.T., 2002. Probabilistic neural network for evaluating seismic liquefaction poten- Karatepe, E., Hiyama, T., 2011. Controlling of artificial neural network for fault diagnosis
tial. Can. Geotech. J. 39 (1), 219–232. of photovoltaic array. In: Intelligent System Application to Power Systems (ISAP),
Gokmen, N., Karatepe, E., Celik, B., Silvestre, S., 2012. Simple diagnostic approach for 16th International Conference, pp. 1–6.
determining of faulted PV modules in string based PV arrays. Sol. Energy 86 (11), Karegar, H.K., Sobhani, B., 2012. Wavelet transform method for islanding detection of
3364–3377. wind turbines. Renewable Energy 38 (1), 94–106.
Gokmen, N., Karatepe, E., Silvestre, S., Celik, B., Ortega, P., 2013. An efficient fault di- Karimi, H., Yazdani, A., Iravani, R., 2008. Negative-sequence current injection for fast
agnosis method for PV systems based on operating voltage-window. Energy Convers. islanding detection of a distributed resource unit. IEEE Trans. Power Electron. 23 (1),
Manage. 73, 350–360. 298–307.
Grimaccia, F., Leva, S., Dolara, A., Aghaei, M., 2017. Survey on PV Modules’ Common Karimi, S., Gaillard, A., Poure, P., Saadate, S., 2008. FPGA-based real-time power con-
Faults After an O & M Flight Extensive Campaign Over Different Plants in Italy. IEEE verter failure diagnosis for wind energy conversion systems. IEEE Trans. Industr.
J. Photovolt. 7 (3), 810–816. Electron. 55 (12), 4299–4308.
Guasch, D., Silvestre, S., Calatayud, R., 2003. Automatic failure detection in photovoltaic Kasemann, M., Grote, D., Walter, B., Kwapil, W., Trupke, T., Augarten, Y., Bardos, R.A.,
systems. In: Photovoltaic Energy Conversion, Proceedings of 3rd World Conference Pink, E., Abbott, M.D., Warta, W., 2008. Luminescence imaging for the detection of
on, vol. 3, pp. 2269–2271. shunts on silicon solar cells. Prog. Photovoltaics Res. Appl. 16 (4), 297–305.
Guerriero, P., d'Alessandro, V., Petrazzuoli, L., Vallone, G., Daliento, S., 2013. Effective Ketola, B., Norris, A., 2011. Degradation mechanism investigation of extended damp heat
real-time performance monitoring and diagnostics of individual panels in PV plants. aged PV modules. 26th EUPVSEC.
In: Clean Electrical Power (ICCEP), International Conference, pp. 14–19. Kim, S.Y., Nam, K., Song, H.S., Kim, H.G., 2008. Fault diagnosis of a ZVS DC–DC con-
Guerriero, P., Piegari, L., Rizzo, R., Daliento, S., 2017. Mismatch Based Diagnosis of PV verter based on DC-link current pulse shapes. IEEE Trans. Industr. Electron. 55 (3),
Fields Relying on Monitored String Currents. Int. J. Photoenergy. 1491–1494.
Hachana, O., Hemsas, K.E., Tina, G.M., 2015. Fault diagnosis of building integrated PV Kirchartz, T., Helbig, A., Reetz, W., Reuter, M., Werner, J.H., Rau, U., 2009. Reciprocity
generator: A metaheuristic technique. In: Renewable Energy Congress (IREC), 6th between electroluminescence and quantum efficiency used for the characterization of
International, pp. 1–6. silicon solar cells. Prog. Photovoltaics Res. Appl. 17 (6), 394–402.
Hachana, O., Tina, G.M., Hemsas, K.E., 2016. PV array fault Diagnostic Technique for Köntges, M., Kurtz, S., Packard, C., Jahn, U., Berger, K.A., Kato, K., Friesen, T., Liu, H.,
BIPV systems. Energy Build. 126, 263–274. Van Iseghem, M., 2014. Review of failures of photovoltaic modules. IEA PVPS Task.
Haeberlin, H., Real, M., 2007. Arc detector for remote detection of dangerous arcs on the Kunte, R.S., Gao, W., 2008. Comparison and review of islanding detection techniques for
DC side of PV plants. In: 22nd EU PV Conf., Milano, Italy. distributed energy resources. In: Power Symposium, NAPS'08. 40th North American,
Haney, J, Burstein, A., 2013. PV System Operations and Maintenance Fundamentals. Next pp. 1–8.
Phase Inc., < http://www.solarabcs.org > . Lahyani, A., Venet, P., Grellet, G., Viverge, P.J., 1998. Failure prediction of electrolytic
Hanif, M., Dwivedi, U.D., Basu, M., Gaughan, K., 2010. Wavelet based islanding detection capacitors during operation of a switchmode power supply. IEEE Trans. Power

183
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

Electron. 13 (6), 1199–1207. on solar photovoltaic array. Environ. Syst. Res. 4 (1), 20.
Lee, S.H., Park, J.W., 2010. New islanding detection method for inverter-based dis- Nie, S., Pei, X., Chen, Y., Kang, Y., 2014. Fault diagnosis of PWM DC–DC converters based
tributed generation considering its switching frequency. IEEE Trans. Ind. Appl. 46 on magnetic component voltages equation. IEEE Trans. Power Electron. 29 (9),
(5), 2089–2098. 4978–4988.
Lezana, P., Aguilera, R., Rodríguez, J., 2009. Fault detection on multi cell converter based Omer, A.M., 2007. Renewable energy resources for electricity generation in Sudan.
on output voltage frequency analysis. IEEE Trans. Industr. Electron. 56 (6), Renew. Sustain. Energy Rev. 11 (7), 1481–1497.
2275–2283. Park, B.G., Lee, K.J., Kim, R.Y., Kim, T.S., Ryu, J.S., Hyun, D.S., 2011. Simple fault di-
Lezana, P., Aguilera, R., Rodríguez, J., 2009. Fault detection on multicell converter based agnosis based on operating characteristic of brushless direct-current motor drives.
on output voltage frequency analysis. IEEE Trans. Industr. Electron. 56 (6), IEEE Trans. Industr. Electron. 58 (5), 1586–1593.
2275–2283. Parzen, E., 1962. On estimation of a probability density function and mode. Ann. Math.
Li, P., Sheng, Y., Zhang, L., Yang, X., Zhao, Y., 2009. A novel active islanding detection Stat. 33 (3), 1065–1076.
method based on current-disturbing. In: Electrical Machines and Systems, 2009. Pei, X., Nie, S., Kang, Y., 2015. Switch short-circuit fault diagnosis and remedial strategy
ICEMS 2009. International Conference 2009, pp. 1–5. for full-bridge DC–DC converters. IEEE Trans. Power Electron. 30 (2), 996–1004.
Li, Z., Wang, Y., Zhou, D., Wu, C., 2012. An intelligent method for fault diagnosis in Peuget, R., Courtine, S., Rognon, J.P., 1998. Fault detection and isolation on a PWM
photovoltaic array. Syst. Simul. Sci. Comput. 10–16. inverter by knowledge-based model. IEEE Trans. Ind. Appl. 34 (6), 1318–1326.
Li, C., Cao, C., Cao, Y., Kuang, Y., Zeng, L., Fang, B., 2014. A review of islanding detection Pham, J.P., Denboer, N., Lidula, N.W., Perera, N., Rajapakse, A.D., 2011. Hardware im-
methods for microgrid. Renew. Sustain. Energy Rev. 35, 211–220. plementation of an islanding detection approach based on current and voltage
Li, B., Shi, S., Wang, B., Wang, G., Wang, W., Xu, D., 2016. Fault diagnosis and tolerant transients. In: Electrical Power and Energy Conference (EPEC), pp. 152–157.
control of single IGBT open-circuit failure in modular multilevel converters. IEEE Pigazo, A., Moreno, V.M., Liserre, M., Dell'Aquila, A., 2007. Wavelet-based islanding
Trans. Power Electron. 31 (4), 3165–3176. detection algorithm for single-phase photovoltaic (PV) distributed generation sys-
Liao, Z., Wang, D., Ren, J., Liu, Z., 2016. An optimized fading fault localization method tems. In: Industrial Electronics, ISIE 2007. IEEE International Symposium, pp.
applied in series-parallel photovoltaic array. In: Smart Grid and Clean Energy 2409–2413.
Technologies (ICSGCE), International Conference, pp. 182–186. Pigazo, A., Liserre, M., Mastromauro, R.A., Moreno, V.M., Dell'Aquila, A., 2009. Wavelet-
Lidula, N.W., Rajapakse, A.D., 2010. A pattern recognition approach for detecting power based islanding detection in grid-connected PV systems. IEEE Trans. Industr.
islands using transient signals—Part I: Design and implementation. IEEE Trans. Electron. 56 (11), 4445–4455.
Power Delivery 25 (4), 3070–3077. Platon, R., Martel, J., Woodruff, N., Chau, T.Y., 2015. Online fault detection in PV sys-
Lin, X., Wang, Y., Zhu, D., Chang, N., Pedram, M., 2012. Online fault detection and tol- tems. IEEE Trans. Sustain. Energy 6 (4), 1200–1207.
erance for photovoltaic energy harvesting systems. In: Proceedings of the Polikar, R., 19993 The story of wavelets. Physics and modern topics in mechanical and
International Conference on Computer-Aided Design, pp. 1–6. electrical engineering, pp. 192–197.
Lin, F.J., Hung, Y.C., Hwang, J.C., Tsai, M.T., 2013. Fault-tolerant control of a six-phase Polo, F.A., del Rosario, J.J., García, G.C., 2010. Supervisory control and automatic failure
motor drive system using a Takagi–Sugeno–Kang type fuzzy neural network with detection in grid-connected photovoltaic systems. In: International Conference on
asymmetric membership function. IEEE Trans. Power Electron. 28 (7), 3557–3572. Industrial, Engineering and Other Applications of Applied Intelligent Systems 2010,
Lin, P., Lin, Y., Chen, Z., Wu, L., Chen, L., Cheng, S., 2017. A density peak-based clus- pp. 458–467.
tering approach for fault diagnosis of photovoltaic arrays. Int. J. Photo Energy. Pu, X.S., Nguyen, T.H., Lee, D.C., Lee, K.B., Kim, J.M., 2013. Fault diagnosis of DC-link
Liu, F., Kang, Y., Zhang, Y., Duan, S., Lin, X., 2010. Improved SMS islanding detection capacitors in three-phase AC/DC PWM converters by online estimation of equivalent
method for grid-connected converters. IET Renew. Power Gener. 4 (1), 36–42. series resistance. IEEE Trans. Industr. Electron. 60 (9), 4118–4127.
Llaria, A., Curea, O., Jimenez, J., Camblong, H., 2009. Survey on Microgrids: analysis of Quintana, E.C., Quintana, M.A., Rolfe, K.D., Thompson, K.R., Hacke, P., 2009. Exploring
technical limitations to carry out new solutions. In: Power Electronics and diagnostic capabilities for application to new photovoltaic technologies. In:
Applications, EPE'09. 13th European Conference, pp. 1–8. Photovoltaic Specialists Conference (PVSC), IEEE, pp. 002031–002036.
Ma, H., Wang, L., 2005. Fault diagnosis and failure prediction of aluminum electrolytic Ray, P.K., Mohanty, S.R., Kishor, N., Dubey, H.C., 2010. Coherency determination in grid-
capacitors in power electronic converters. In: Industrial Electronics Society, IECON, connected distributed generation based hybrid system under islanding scenarios. In:
31st Annual Conference of IEEE, p.6. Power and Energy (PECon), IEEE International Conference, pp. 85–88.
Madeti, S.R., Singh, S.N., 2017. Comparative analysis of solar photovoltaic monitoring Ray, P.K., Kishor, N., Mohanty, S.R., 2010. S-transform based islanding detection in grid-
systems. AIP Conf. Proc. 1859 (1), 020116. connected distributed generation based power system. In: Energy Conference and
Madeti, S.R., Singh, S.N., 2017a. Monitoring system for photovoltaic plants: A review. Exhibition (EnergyCon), IEEE International Conference, pp. 612–617.
Renew. Sustain. Energy Rev. 67, 1180–1207. Ray, P.K., Mohanty, S.R., Kishor, N., 2011. Disturbance detection in grid-connected dis-
Madeti, S.R., Singh, S.N., 2017b. Online fault detection and the economical analysis of tributed generation system using wavelet and S-transform. Electr. Power Syst. Res. 81
grid connected photovoltaic systems. Energy 134, 121–135. (3), 805–819.
Mahat, P., Chen, Z., Bak-Jensen, B., 2008. Review of islanding detection methods for Rezgui, W., Mouss, H., Mouss, N., Mouss, D., Benbouzid, M., Amirat, Y., 2015.
distributed generation. In: Electric Utility Deregulation and Restructuring and Power Photovoltaic module simultaneous open-and short-circuit faults modeling and de-
Technologies, DRPT Third International Conference, pp. 2743–2748. tection using the I-V characteristic. In: Industrial Electronics (ISIE), IEEE 24th
Mahat, P., Chen, Z., Bak-Jensen, B., 2009. A hybrid islanding detection technique using International Symposium, pp. 855–860.
average rate of voltage change and real power shift. IEEE Trans. Power Delivery 24 Ribeiro, E., Cardoso, A.J., Boccaletti, C., 2013. Fault-tolerant strategy for a photovoltaic
(2), 764–771. DC–DC converter. IEEE Trans. Power Electron. 28 (6), 3008–3018.
Mahat, P., Chen, Z., Bak-Jensen, B., 2011. Review on islanding operation of distribution Ribeiro, E., Cardoso, A.J., Boccaletti, C., 2014. Open-circuit fault diagnosis in interleaved
system with distributed generation. In: Power and Energy Society General Meeting, DC–DC converters. IEEE Trans. Power Electron. 29 (6), 3091–3102.
pp. 1–8. Rodríguez-Blanco, M.A., Claudio-Sánchez, A., Theilliol, D., Vela-Valdés, L.G., Sibaja-
Masrur, M.A., Chen, Z., Murphey, Y., 2010. Intelligent diagnosis of open and short circuit Terán, P., Hernández-González, L., Aguayo-Alquicira, J., 2011. A failure-detection
faults in electric drive inverters for real-time applications. IET Power Electron. 3 (2), strategy for IGBT based on gate-voltage behavior applied to a motor drive system.
279–291. IEEE Trans. Industr. Electron. 58 (5), 1625–1633.
Meinguet, F., Sandulescu, P., Kestelyn, X., Semail, E., 2013. A method for fault detection Ropp, M., Ginn, J., Stevens, J., Bower, W., Gonzalez, S., 2006. Simulation and experi-
and isolation based on the processing of multiple diagnostic indices: application to mental study of the impedance detection anti-islanding method in the single-inverter
inverter faults in AC drives. IEEE Trans. Veh. Technol. 62 (3), 995–1009. case. In: Photovoltaic Energy Conversion, Conference Record of the IEEE 4th World
Mekki, H., Mellit, A., Salhi, H., 2016. Artificial neural network-based modelling and fault Conference, vol. 2, pp. 2379–2382.
detection of partial shaded photovoltaic modules. Simul. Model. Pract. Theory Sabbaghpur Arani, M., Hejazi, M.A., 2016. The Comprehensive Study of Electrical Faults
67, 1–3. in PV Arrays. J. Electr. Comput. Eng. 2016.
Menon, V., Nehrir, M.H., 2007. A hybrid islanding detection technique using voltage Salehifar, M., Arashloo, R.S., Moreno-Equilaz, J.M., Sala, V., Romeral, L., 2014. Fault
unbalance and frequency set point. IEEE Trans. Power Syst. 22 (1), 442–448. detection and fault tolerant operation of a five phase PM motor drive using adaptive
Miwa, M., Yamanaka, S., Kawamura, H., Ohno, H., 2006. Diagnosis of a power output model identification approach. IEEE J. Emerg. Selected Topics Power Electron. 2 (2),
lowering of PV array with a (−dI/dV)-V characteristic. In: Photovoltaic Energy 212–223.
Conversion, Conference Record of the IEEE 4th World Conference, vol. 2, pp. Samantaray, S.R., Pujhari, T.M., Subudhi, B.D., 2009. A new approach to islanding de-
442–2445. tection in distributed generations. In: Power Systems, 2009. ICPS'09. International
Moradzadeh, M., Rajabzadeh, M., Bathaee, S.M., 2008. A novel hybrid islanding detection Conference on 2009, pp. 1–6.
method for distributed generations. In: Electric Utility Deregulation and Samantaray, S.R., Samui, A., Babu, B.C., 2010. S-transform based cumulative sum de-
Restructuring and Power Technologies, 2008. DRPT 2008. Third International tector (CUSUM) for islanding detection in Distributed Generations. In: Power
Conference on, pp. 2290–2295. Electronics, Drives and Energy Systems (PEDES) & 2010 Power India, 2010 Joint
Moreno-Garcia, Isabel M., et al., 2016. Real-time monitoring system for a utility-scale International Conference, pp. 1–6.
photovoltaic power plant. Sensors 16 (6), 770. Samantaray, S.R., Babu, B.C., Dash, P.K., 2011. Probabilistic neural network based is-
Morsi, W.G., Diduch, C.P., Chang, L., 2010. A new islanding detection approach using landing detection in distributed generation. Electr. Power Compon. Syst. 39 (3),
wavelet packet transform for wind-based distributed generation. In: Power 191–203.
Electronics for Distributed Generation Systems (PEDG), 2010 2nd IEEE International Schapire, R., 2008. Theoretical Machine Learning. Computer Science lecture. Princeton
Symposium, pp. 495–500. University, New Jersey, United States.
Mukaro, R., Tinarwo, D., 2008. Performance evaluation of a hot-box reflector solar cooker Schimpf, F., Norum, L.E., 2009. Recognition of electric arcing in the DC-wiring of pho-
using a microcontroller-based measurement system. Int. J. Energy Res. 32 (14), tovoltaic systems. In: Telecommunications Energy (INTELEC) International IEEE
1339–1348. Conference, pp. 1–6.
Nguyen, X.H., 2015. Matlab/Simulink based modeling to study effect of partial shadow Schirone, L., Califano, F.P., Pastena, M., 1994. Fault detection in a photovoltaic plant by

184
S.R. Madeti, S.N. Singh Solar Energy 158 (2017) 161–185

time domain reflectometry. Prog. Photovoltaics Res. Appl. 2 (1), 35–44. thermal effect of defects on silicon-based PV cells. IEEE Trans. Industr. Electron. 59
Senroy, N., Heydt, G.T., Vittal, V., 2006. Decision tree assisted controlled islanding. IEEE (10), 3860–3867.
Trans. Power Syst. 21 (4), 1790–1797. Wang, E., Yen, K.H., Wang, C., Ji, L., Zgonena, T., 2011. Accelerated aging tests on PV
Seo G.S., Bae H, Cho B.H., Lee K.C., 2012. Arc protection scheme for DC distribution grounding connections. Energy Procedia 12, 578–585.
systems with photovoltaic generation. In: Renewable Energy Research and Wolgemuth, J., 2010. Failure modes of crystalline Si modules’, PV Module Reliability
Applications (ICRERA), IEEE International Conference, pp. 1–5. Workshop.
Shahbazi, M., Jamshidpour, E., Poure, P., Saadate, S., Zolghadri, M.R., 2013. Open-and Wu, Y., Lan, Q., Sun, Y., 2009. Application of BP neural network fault diagnosis in solar
short-circuit switch fault diagnosis for nonisolated dc–dc converters using field pro- photovoltaic system. In: Mechatronics and Automation, International Conference on
grammable gate array. IEEE Trans. Industr. Electron. 60 (9), 4136–4146. 2009, pp. 2581–2585.
Shao, S., Wheeler, P.W., Clare, J.C., Watson, A.J., 2013. Fault detection for modular Xu, X., Wang, H., Zuo, Y., 2011. Method for diagnosing photovoltaic array fault in solar
multilevel converters based on sliding mode observer. IEEE Trans. Power Electron. 28 photovoltaic system. In: Power and Energy Engineering Conference (APPEEC),
(11), 4867–4872. pp. 1–5.
Shariatinasab, R., Akbari, M., 2010. New islanding detection technique for DG using Yafaoui, A., Wu, B., Kouro, S., 2012. Improved active frequency drift anti-islanding de-
discrete wavelet transform. In: Power and Energy (PECon), 2010 IEEE International tection method for grid connected photovoltaic systems. IEEE Trans. Power Electron.
Conference, pp. 294–299. 27 (5), 2367–2375.
Sharma, V., Chandel, S.S., 2013. Performance and degradation analysis for long term Yagi, Y., Kishi, H., Hagihara, R., Tanaka, T., Kozuma, S., Ishida, T., Waki, M., Tanaka, M.,
reliability of solar photovoltaic systems: a review. Renew. Sustain. Energy Rev. 27, Kiyama, S., 2003. Diagnostic technology and an expert system for photovoltaic sys-
753–767. tems using the learning method. Sol. Energy Mater. Sol. Cells 75 (3), 655–663.
Silvestre, S., Chouder, A., Karatepe, E., 2013. Automatic fault detection in grid connected Yao, X., Herrera, L., Ji, S., Zou, K., Wang, J., 2014. Characteristic study and time-domain
PV systems. Sol. Energy 94, 119–127. discrete-wavelet-transform based hybrid detection of series DC arc faults. IEEE Trans.
Silvestre, S., da Silva, M.A., Chouder, A., Guasch, D., Karatepe, E., 2014. New procedure Power Electron. 29 (6), 3103–3115.
for fault detection in grid connected PV systems based on the evaluation of current Yazdani, A., Sepahvand, H., Crow, M.L., Ferdowsi, M., 2011. Fault detection and miti-
and voltage indicators. Energy Convers. Manage. 86, 241–249. gation in multilevel converter STATCOMs. IEEE Trans. Industr. Electron. 58 (4),
Silvestre, S., Kichou, S., Chouder, A., Nofuentes, G., Karatepe, E., 2015. Analysis of cur- 1307–1315.
rent and voltage indicators in grid connected PV (photovoltaic) systems working in Yin, G., 2005. A distributed generation islanding detection method based on artificial
faulty and partial shading conditions. Energy 86, 42–50. immune system. In: Transmission and Distribution Conference and Exhibition: Asia
Silvestre, S., Mora-López, L., Kichou, S., Sánchez-Pacheco, F., Dominguez-Pumar, M., and Pacific, IEEE/PES, pp. 1–4.
2016. Remote supervision and fault detection on OPC monitored PV systems. Sol. Yoo, C.H., Jang, D.H., Han, S.K., Oh, D.S., Hong, S.S., 2011. A new phase drift anti-
Energy 137, 424–433. islanding method for grid-connected inverter system. In: Power Electronics and ECCE
Sleszynski, W., Nieznanski, J., Cichowski, A., 2009. Open-transistor fault diagnostics in Asia (ICPE & ECCE), 2011 IEEE 8th International Conference on 2011, pp. 902–906.
voltage-source inverters by analyzing the load currents. IEEE Trans. Industr. Electron. Youssef, A.B., El Khil, S.K., Slama-Belkhodja, I., 2013. State observer-based sensor fault
56 (11), 4681–4688. detection and isolation, and fault tolerant control of a single-phase PWM rectifier for
Smith, K.S., Ran, L., Penman, J., 1997. Real-time detection of intermittent misfiring in a electric railway traction. IEEE Trans. Power Electron. 28 (12), 5842–5853.
voltage-fed PWM inverter induction-motor drive. IEEE Trans. Industr. Electron. 44 Yu, G.J., So, J.H., Jung, Y.S., Choi, J.Y., Jeong, S.G., Kim, K.H., Lee, K.O., 2005. Boundary
(4), 468–476. conditions of reactive-power-variation method and active-frequency-drift method for
So, J.H., Jung, Y.S., Yu, G.J., Choi, J.Y., Choi, J.H., 2007. Performance results and ana- islanding detection of grid-connected photovoltaic inverters. In: Photovoltaic
lysis of 3kW grid-connected PV systems. Renewable Energy 32 (11), 1858–1872. Specialists Conference Record of the Thirty-first IEEE, pp. 1785–1787.
Solórzano, J., Egido, M.A., 2013. Automatic fault diagnosis in PV systems with distributed Yu, B.G., Matsui, M., Yu, G.J., 2011. A correlation-based islanding-detection method
MPPT. Energy Convers. Manage. 76, 925–934. using current-magnitude disturbance for PV system. IEEE Trans. Industr. Electron. 58
Standard IE, 1995. 61215 Crystalline Silicon Terrestrial Photovoltaic (PV) ModulesDesign (7), 2935–2943.
Qualification and Type Approval. International Electrotechnical Commission. Zeineldin, H.H., Kirtley, J.L., 2009 April Apr. A simple technique for islanding detection
Stellbogen, D., 1993. Use of PV circuit simulation for fault detection in PV array fields. In: with negligible non detection zone. IEEE Trans. Power Delivery 24 (2), 779–786.
Photovoltaic Specialists Conference Record of the Twenty Third IEEE, pp. Zeineldin, H.H., Salama, M.M., 2011. Impact of load frequency dependence on the NDZ
1302–1307. and performance of the SFS islanding detection method. IEEE Trans. Industr.
Takashima, T., Yamaguchi, J., Otani, K., Kato, K., Ishida, M., 2006. Experimental studies Electron. 58 (1), 139–146.
of failure detection methods in PV module strings. In: Photovoltaic Energy Zeineldin, H.H., El-Saadany, E.F., Salama, M.M., 2006. Impact of DG interface control on
Conversion, Conference Record of the IEEE 4th World Conference, vol. 2, pp. islanding detection and non detection zones. IEEE Trans. Power Delivery 21 (3),
2227–2230. 1515–1523.
Takashima, T., Yamaguchi, J., Ishida, M., 2008. Fault detection by signal response in PV Zhao, Y., 2010. Fault Analysis in Solar Photovoltaic Arrays. Northeastern University,
module strings. In: Photovoltaic Specialists Conference, 33rd IEEE, pp. 1–5. Boston, Mass, USA.
Takashima, T., Yamaguchi, J., Otani, K., Oozeki, T., Kato, K., Ishida, M., 2009. Zhao, Y., Lehman, B., de Palma, J.F., Mosesian, J., Lyons, R., 2011. Challenges to over
Experimental studies of fault location in PV module strings. Sol. Energy Mater. Sol. current protection devices under line-line faults in solar photovoltaic arrays. In:
Cells 93 (6), 1079–1082. Energy Conversion Congress and Exposition (ECCE), IEEE, pp. 20–27.
Task, V., 2002. Probability of islanding in grid networks due to grid-connected photo- Zhao, Y., Lehman, B., de Palma, J.F., Mosesian, J., Lyons, R., 2011. Fault analysis in solar
voltaic power systems. Tech. Rep. IEA-PVPS T5-07. PV arrays under: Low irradiance conditions and reverse connections. In: Photovoltaic
Timbus, A., Oudalov, A., Ho, C.N., 2010. Islanding detection in smart grids. In: Energy Specialists Conference (PVSC), 37th IEEE, pp. 002000–002005.
Conversion Congress and Exposition (ECCE), pp. 3631–3637. Zhao, Y., Yang, L., Lehman, B., de Palma, J.F., Mosesian, J., Lyons, R., 2012. Decision
Vahedi, H., Noroozian, R., Jalilvand, A., Gharehpetian, G.B., 2010. Hybrid SFS and Q-f tree-based fault detection and classification in solar photovoltaic arrays. In: Applied
Islanding Detection Method for inverter-based DG. In: Power and Energy (PECon), Power Electronics Conference and Exposition (APEC), Twenty-Seventh Annual IEEE,
IEEE International Conference, pp. 672–676. pp. 93–99.
Veldman, D., Bennett, I.J., Brockholz, B., de Jong, P.C., 2011. Non-destructive testing of Zhao, Y., Lehman, B., Ball, R., Mosesian, J., de Palma, J.F., 2013. Outlier detection rules
crystalline silicon photovoltaic back-contact modules. In: Photovoltaic Specialists for fault detection in solar photovoltaic arrays. In: Applied Power Electronics
Conference (PVSC), 2011 37th IEEE, pp. 003237–003240). Conference and Exposition (APEC), Twenty-Eighth Annual IEEE, pp. 2913–2920.
Ventura, C., Tina, G.M., 2016. Utility scale photovoltaic plant indices and models for on- Zhao, Y., Ball, R., Mosesian, J., de Palma, J.F., Lehman, B., 2015. Graph-based semi-
line monitoring and fault detection purposes. Electr. Power Syst. Res. 136, 43–56. supervised learning for fault detection and classification in solar photovoltaic arrays.
Vergura, S., Acciani, G., Amoruso, V., Patrono, G., 2008. Inferential statistics for mon- IEEE Trans. Power Electron. 30 (5), 2848–2858.
itoring and fault forecasting of PV plants. In: Industrial Electronics, IEEE Zhu, Y., Yang, Q., Wu, J., Zheng, D., Tian, Y., 2008. A novel islanding detection method of
International Symposium on, pp. 2414–2419. distributed generator based on wavelet transform. In: Electrical Machines and
Vergura, S., Acciani, G., Falcone, O., 2012. A finite-element approach to analyze the Systems, 2008. ICEMS 2008. International Conference on, pp. 2686–2688.

185