Advanced Fault Detection Technique for AC

Microgrid Protection
Rudranarayan Pradhan Premalata Jena
Department of Electrical Engineering Department of Electrical Engineering
Odisha University of Technology and Research, IIT Roorkee
Bhubaneswar, Odisha Uttarakhand, India
2023 IEEE 3rd International Conference on Sustainable Energy and Future Electric Transportation (SEFET) | 979-8-3503-1997-2/23/$31.00 ©2023 IEEE | DOI: 10.1109/SEFET57834.2023.10245640

[email protected] [email protected]

Abstract— Microgrid (MG) plays a crucial role in The primary requirement of MG protection is fault
developing the modern smart grid. A microgrid can be detection. Efficient techniques based on recent findings
integrated with renewable energy sources as a practical way have been proposed by many researchers. The most popular
to guarantee consumers' supply security. Designing an methods of fault detection are
appropriate protection mechanism is one of the key technical
challenges in implementing a MG. The technique must 1. Using the voltage waveform features
operate in grid-following and grid-isolated modes while still 2. Voltage and current symmetrical component.
achieving the fundamental protection needs of selectivity,
3. Differential values
reliability, and sensitivity. Traditional protection solutions
are inappropriate for MG due to bidirectional current flow, 4. ML-based techniques, synchro phasors, data-
small fault currents in the inverter interface MG, and mining, multi-agents, internet of things, etc.
variations in their operation methods. Due to the change in
the operation method of the MG, the magnitude and direction In distortion in voltage waveform, network voltage drop
of the fault current are also reversed, for which conventional and harmonics distortion of the current waveform are
overcurrent protection systems perform poorly. For the MG proposed in [4],[5]. However, voltage and current
to operate reliably, the fault must be identified and fixed as waveform may not be distorted in all fault conditions.
soon as possible as it appears. This paper suggests a fault
detection approach for AC MG using the Transient A sequence component-based protection strategy is
Monitoring Function Index (TMFI) in a grid-connected mode suggested to determine grounding faults in MG [6]. A
that detects the fault efficiently. To assess the effectiveness of negative sequence current is used to determine
the recommended methodology, a test IEEE seven-bus MG unsymmetrical fault in MG. However, the following
system is modeled in the Real-Time Digital Simulation method is not suitable for determining the balanced fault.
(RTDS) framework, and the suggested strategy is evaluated.
The interpretation was interpreted in the MATLAB A differential quantity-based technique is proposed [7],
environment. The simulation results confirm the accuracy of [8]. These techniques are most efficient in determining the
the suggested technique. fault in the MG and can accurately detect the fault in the
MG. However, these techniques require additional
Keywords— MG, fault detection, Faults, TMFI, PV, DFIG.
communication networks for proper operation.
I. INTRODUCTION In [9], the authors suggest a protective strategy using
Concerns about climate change are driving the switch inverter control and microprocessor-relay (McR)
from traditional fossil fuels to renewable energy sources parameters. In order that McR can precisely identify the
(RES) in the power generation industry. As power demand problem, the system uses software current control to inject
continues to rise, distributed energy resources (DERs), like a portion of the 5th harmonic into the fault current. By
photovoltaics (PVs), wind turbines, fuel cells, and restricting currents with hardware hysteresis and current
microturbines, are increasingly being adopted due to their software control, the inverter is additionally safeguarded
various advantages such as reduced CO2 emissions, energy from short-circuit problems.
cost savings, reduced line costs, decreased transmission In [10], an effective ML-based adaptive protection
losses, and improved system reliability. However, the system is described. The plan initially analyses uncertain
integration of these DERs into the power network can pose elements using Pearson correlation coefficients from data
certain challenges that need to be addressed to ensure the mining, followed by MG state detection using a hybrid
smooth operation of the MG. To address these challenges, artificial neural network (HANN) and support vector
a directional overcurrent relay (DOCR) based protection machines (SVM) model based on big data streams. The
scheme for the MG has been developed, which is a critical suggested approach also establishes network
technique for safeguarding such systems. To make sure the reconfigurations and adaptive protective settings.
reliable and accurate operation of DERs in the microgrid, it
is imperative to analyze and resolve any protection issues. For the classification of faults in MG, a novel approach
This is because, primarily employing a current limiting that relies on a semi-supervised ML strategy with the
method built into the inverter control system, the inverter training of k- nearest neighbors and decision trees are put
current is normally restricted during short-circuit faults to forth in [11]. Because of the utilization of labeled and
2-3 times the rated current [1]–[3]. unlabelled data for developing and training the classifier
model, it performs better than earlier supervised ML

979-8-3503-1997-2/23/$31.00©2023 IEEE
techniques. ML-based techniques provide accurate results
but require a large number of data for analysis.
B. Proposed Technique
In [12], a differential protection strategy using data
This research suggests a TMFI-based fault detection
mining is proposed. To estimate the sensitive features, the
technique in DG interfaced with the MG. Initially the
voltage and current samples are analyzed throughout the
voltage sample obtained at the relay location is analyzed
fault. Then, a decision tree-based data-mining model is
through the FFT. The TMF is the variation between the
built using these features to compute the differential
estimated current sample generated, and actual current. At
characteristics, which are then applied to generate the final
the end TMF index (TMFI) is calculated by averaging the
relay output.
TMF index of all phases.
However, a definite investigation of MG protection plans
based on conventional protection tenets and unique ways is
still missing.
The remaining portion of the research is outlined as
follows. Test IEEE seven bus MG and details of the
suggested technique are described in Section II. Section III
analyses the result, and Section IV includes the discussion
and final thoughts.
II. PROPOSED TECHNIQUE
A. System Description
To verify the proposed technique, an IEEE modified
seven bus (IMSB) medium voltage MG model is simulated
in RTDS software as represented in Fig. 1. The IMSB
system is the most appropriate for the research since it
offers a wide range of features, good topological
characteristics, and multiple long feeders for integrating
various renewable sources for rural distribution network
research. The standard IMSB is modified by adding solar
PV (1.74 MW) at bus B3 and a DFIG-based wind turbine
(DWT) of 2 MW system at bus B5, respectively. A diesel
generator (3 MW diesel engine-based synchronous
machine) is connected to bus B7. The PV and DFIG energy
sources are interfaced with the AC MG grid using switched
VSC models with sinusoidal PWM controls. A switched Fig. 1. Seven Bus Test Microgrid
capacitor bank (0.5 MVAr) is used to balance and control
the voltages at the point of connection bus B1. The grid is
operated at 138 kV at a frequency of 60 Hz. To optimize
the power output of a photovoltaic (PV) array, a maximum
power point tracker is employed in the PV control system,
which maintains the DC voltage at the full power point. A
decoupled current control system is utilized to regulate
the exchange of power between the PV system and the grid.
This enables precise control of the power flow in the
system, enhancing its efficiency and stability. By
implementing these control techniques, the performance of
the PV system can be maximized, contributing to increased
renewable energy utilization and reduced environmental
impact. In this model, MG is modeled radial network. The
following criteria must be followed for grid
synchronization after the island operation, as given in
Table: I
TABLE: I
Fig. 2. Moving Data window

Frequency Phase ( Voltage Total DER Fig. 2 displays a moving sample current data window for
( f )  ) ( V ) the waveform at the relay location prior to and during a
fault state, for which TMF is estimated. The least-square
0.1 Hz 100 3% 1.5×103- estimation (LSE) approach is utilized to estimate the
10×103kVA fundamental frequency component of the current
waveform in the first stage. Under typical circumstances,
 C = [i(t0 + N s ) i(t0 + 2 N s ) ....
(5)
...i (t0 + KN s )]T
Where k denotes the quantity of samples in a cycle. The
fundamental component of current is determined using
LSE method.

C = X 1 P1 (6)

Where P1 = [u1 , u2 ]T
Fig. 3. Reconstructed and actual current wave

the estimated basic component's reconstruction and the T

actual signal are matched as shown in Fig. 3. The inverter cos(0 N s ) cos(0 2 N s ) . . cos(0 KTs ) 
X1 =  
current changes instantly when a fault happens, deviating
 sin(0 N s ) sin(0 2 N s ) . . sin(0 KTs ) 
from the signal that has been reconstructed. Finally, by
specifying a suitable TMF, the change from normal (7)
operating condition to faulty operating conditions can be
accurately predicted [13]. The LS solution P1 for the estimation of the coefficient
is [13]
Let the current signal is processed as
Pˆ1 = ( X 1T X 1 )−1 X 1T C (8)
2K
i (t ) =  uk X k (t ) (1) Using eqn (8), the reconstruction of current samples C
k
can be determined from the estimate P̂1 as

C = X1Pˆ1 = X 1 ( X 1T X 1 )−1 X 1T C (9)

where k denotes the signal's highest harmonic order
X k (t ) are known signal If C  C , it indicates deviation from the normal
conditions, which indicates a fault.
u k are known coefficients
The mismatch can be calculated as
The popular choices for the signal X k (t ) are
Q = C − C = [ X 1 ( X 1T X 1 ) −1 X 1T − I ]C
T T (10)
 X1   cos(0t ) 
X   sin( t ) 
 2  0  Where the residual is Q = [q1 , q2, q3 ....qn ]T
X =  X 3  = cos(20t )  (2)
   
 .   .  The TMF is defined as the sum of absolute values of qk
 .   .  in a cycle, i.e.,

Where 0 is the fundamental frequency (f) of the

N
current signal TMF =  qn (11)
i (t ) = u1 cos(0t ) + u2 sin(0t ) + n =1

(3)
u3 cos(20t ) + u4 sin(20t ) + ....
The change from normal to fault situation can be
First, with a sampling period N s , the signal is sampled. calculated by obtaining the max value of TMF among all
phases.
The discrete form is represented as
2K
TMFI=max(TMFR ,TMFY ,TMFB ) (12)
cn =  uk X k ( nTs ) (4) A fault condition is detected when
k
TMFI> D TH (13)
Where cn is the nth element of the current
measurement vector C
 DTH indicates is the threshold decided earlier by destemming the fault instant. By suitable choosing the
value of DTH the fault can be accurately detected. The
considering various conditions such as fault resistance and
proposed strategy flow chat is provided in Fig. 4.
operating voltage etc.
Fig . 5. represents a L-G fault simulated between the R
phase and the ground of the proposed MG at 0.2 sec.
The L-L fault between R –Y phase of the proposed MG
is created at 0.2 sec shown Fig. 6 . The system is allowed
to run for 1 sec.

Fig. 6. L-L Fault at R-Y phase

The TMFI (Index of TMF) is calculated for LG fault of

the proposed MG represented in Fig. 7. The threshold value
(DTH) is determined by considering various factors such as
fault resistance, load change, system voltage etc. When the
TMFI exceeds the DTH, a triggered signal is generated by the
Fig. 4. Flow chart of the proposed methodology relay, and the fault is isolated.
The L-L fault incepted between the R and Y phase of the
suggested MG is calculated. When the TMFI value exceeds
III. RESULTS AND DISCUSSION the DTH, the relay provides a trip signal to the CB to isolate
the fault as shown in Fig. 8.
To analyze the recommended method, an IEEE seven bus
system test MG is modelled in an RTDS environment.
Numerous types of symmetrical and unsymmetrical faults
are simulated at different points of the proposed MG to
confirm the effectiveness of the proposed strategy.
The capability to connect physical protection relays in
order to test their functionality in actual microgrid
operating environment is the main advantage of RTDS-
based real-time simulation.
The results are analyzed in the MATLAB environment.
The MG system is allowed to function at 60 Hz frequency;
the sampling frequency is taken as 6 kHz i.e., 100
samples/cycle is taken to get a better accuracy of the
suggested methodology. The fault data are collected at the Fig. 7. TMFI for L-G
relay location. DTH (threshold) plays a vital role in

Fig. 8. TMFI for L-L fault

Fig. 5. LG Fault at R phase
 The TMFI for the 3-Φ symmetrical fault is calculated by
using equation (12). The DTH is determined for the above
fault and it is compared with TMFI. When the TMFI > DTH
trip signal is produced by the relay in the designed MG and
the faulty section is isolated as represented in Fig. 10.
The pole-pole-Pole-ground fault is created in the
proposed MG. The magnitude of the fault current rises
significantly from the normal value, as shown in Fig.11.
The TMFI for the above fault is calculated shown in Fig.
12. When the TMFI index exceeds the predefined value of
the threshold, the relay generates the trip signal, and the
faulty section is isolated by the circuit breaker.
Fig. 9. L-L-L Fault Current Wave

IV. CONCLUSIONS
Conventional protection schemes fail to detect the
Microgrid (MG) fault. For efficient operation of MG the
fault should be identified within a possible minimum time.
This research proposes an efficient technique based on
Transient Monitoring Function Index (TMFI), which can
detect symmetrical and unbalanced fault effectively. To
verify the proposed technique a seven bus test MG system
is simulated in a Real-time environment of the RTDS
platform to provide a realistic simulation of the
microprocessor-based physical relays. This technique
Fig. 10. TMFI for L-L-L Fault
doesn’t require any communication network, reducing its
cost to implement in microprocessor-based relays.

REFERENCES
[1] R. Pradhan and P. Jena, “An innovative fault
direction estimation technique for AC microgrid,”
Electr. Power Syst. Res., vol. 215, no. PA, p.
108997, 2023, doi: 10.1016/j.epsr.2022.108997.
[2] N. K. Sharma and S. R. Samantaray, “PMU
Assisted Integrated Impedance Angle-Based
Fig. 11. Fault Current for L-L–L–G fault
Microgrid Protection Scheme,” IEEE Trans.
Power Deliv., vol. 35, no. 1, pp. 183–193, 2020,
doi: 10.1109/TPWRD.2019.2925887.
[3] R. Pradhan and P. Jena, “Comparison of Phasor
Estimation Techniques for AC Microgrid
Protection,” 2021 Int. Symp. Asian Control Assoc.
Intell. Robot. Ind. Autom. IRIA 2021, no.
September, pp. 31–36, 2021, doi:
10.1109/IRIA53009.2021.9588709.
[4] A. B. Nassif, “A Protection and Grounding
Strategy for Integrating Inverter-Based Distributed
Energy Resources in an Isolated Microgrid,” CPSS
Trans. Power Electron. Appl., vol. 5, no. 3, pp.
Fig. 12. TMFI L-L–L–G fault
242–250, 2020, doi:
10.24295/cpsstpea.2020.00021.
[5] S. M. Brahma, “Fault location in power distribution
The Pole-Pole-Pole (L-L-L) fault among R-Y-B phases system with penetration of distributed generation,”
of the MG is created. The fault duration is kept for 0.15 IEEE Trans. Power Deliv., vol. 26, no. 3, pp. 1545–
sec. Once the fault is triggered, the magnitudes of the 1553, 2011, doi: 10.1109/TPWRD.2011.2106146.
current increase from the normal value significantly, as
shown in the Fig. 9. [6] A. Hooshyar and R. Iravani, “Microgrid
 Protection,” Proc. IEEE, vol. 105, no. 7, pp. 1332–
1353, 2017, doi: 10.1109/JPROC.2017.2669342.
[7] K. Dubey and P. Jena, “Impedance Angle-Based
Differential Protection Scheme for Microgrid
Feeders,” IEEE Syst. J., vol. 15, no. 3, pp. 3291–
3300, Sep. 2021, doi:
10.1109/JSYST.2020.3005645.
[8] E. Casagrande, W. L. Woon, H. H. Zeineldin, and
D. Svetinovic, “A differential sequence component
protection scheme for microgrids with inverter-
based distributed generators,” IEEE Trans. Smart
Grid, vol. 5, no. 1, pp. 29–37, 2014, doi:
10.1109/TSG.2013.2251017.
[9] Z. Chen, X. Pei, M. Yang, L. Peng, and P. Shi, “A
Novel Protection Scheme for Inverter-Interfaced
Microgrid (IIM) Operated in Islanded Mode,”
IEEE Trans. Power Electron., vol. 33, no. 9, pp.
7684–7697, 2018, doi:
10.1109/TPEL.2017.2769559.
[10] H. Lin, K. Sun, Z. H. Tan, C. Liu, J. M. Guerrero,
and J. C. Vasquez, “Adaptive protection combined
with machine learning for microgrids,” IET Gener.
Transm. Distrib., vol. 13, no. 6, pp. 770–779, 2019,
doi: 10.1049/iet-gtd.2018.6230.
[11] T. S. Abdelgayed, W. G. Morsi, and T. S. Sidhu,
“Fault detection and classification based on co-
training of semisupervised machine learning,”
IEEE Trans. Ind. Electron., vol. 65, no. 2, pp.
1595–1605, 2017, doi:
10.1109/TIE.2017.2726961.
[12] S. Kar, S. R. Samantaray, and M. D. Zadeh, “Data-
Mining Model Based Intelligent Differential
Microgrid Protection Scheme,” IEEE Syst. J., vol.
11, no. 2, pp. 1161–1169, 2017, doi:
10.1109/JSYST.2014.2380432.
[13] I. Sadeghkhani, M. E. H. Golshan, A. Mehrizi-
Sani, J. M. Guerrero, and A. Ketabi, “Transient
Monitoring Function-Based Fault Detection for
Inverter-Interfaced Microgrids,” IEEE Trans.
Smart Grid, vol. 9, no. 3, pp. 2097–2107, May
2018, doi: 10.1109/TSG.2016.2606519.