IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL.

73, 2024 3527712

Amplitude-Phase-Characteristics-Based Fault Phase

Detection Method for Grounding Fault
in Distribution Network
Jiahao Lin , Student Member, IEEE, Jianzhang You , Member, IEEE, Moufa Guo , Member, IEEE,
and Qiteng Hong , Senior Member, IEEE

Abstract— The fault phase detection (FPD) of single-line-to- percentage of access to active devices and power electronic
ground fault (SLGF) is important for active arc suppression components leads to an increase in the active and harmonic
devices (AASD), but it is challenging when considering asym- components of the fault current [5]. Therefore, it is difficult
metrical line parameters and different neutral grounding types.
To improve the reliability of FPD, this article proposes an
for the traditional passive arc suppression method to suppress
FPD method based on the fusion of amplitude and phase angle the fault current.
characteristics. First, when an SLGF occurs, the amplitude and The active arc suppression technology refers to that the fault
phase angle characteristics of the three-phase source voltage are current is suppressed by means of active intervention, which
analyzed. Then, the phase angle of the fault phase source voltage includes phase Earth systems (PESs), ground-fault neutral-
is calculated based on the derived amplitude and phase angle izer (GFN), cascaded-H-bridge (CHB)-based arc suppression
characteristics, which are compared with the measured phase
methods, etc., [6], [7], [8], [9], [10]. Among them, the fault
angle of the three-phase source voltage, respectively. According to
the FPD criterion, the fault phase can be determined. In addition, phase voltage is suppressed by PES by diverting the fault
an FPD device (FPDD) featuring convenient installation has been current to the ground. The structure of PES is simple and
designed. The FPD results are sent to AASD, if the fault phase relatively easy to implement. However, the load of the feeders
voltage can be effectively suppressed, the FPD results are correct. may affect the effectiveness of arc suppression [11]. During
The proposed method has been experimentally validated on 380 V the arc suppression of GFN, the capacitive component of the
and 10 kV experimental platforms. fault current is compensated by an arc suppression coil (ASC),
Index Terms— Active arc suppression device (AASD), asym- and the fault residual current is compensated by a single-
metric line parameters, distribution network, fault phase phase inverter. The effectiveness of fault current suppression is
detection (FPD), single-line-to-ground fault (SLGF). superior to ASC. However, this single-cascade configuration
may result in a high harmonic content of the compensation
I. I NTRODUCTION current [12]. Wang et al. [10] and Fan et al. [13] proposed a

D ISTRIBUTION networks are the last link between elec-

tric utilities and customers, which play an important role
in keeping a reliable and safe power supply. Single-line-to-
fault-phase voltage active regulation strategy for transformer
LV-side access inverter, which effectively suppressed the fault
current, and the practicality of the strategy was verified
ground faults (SLGFs) account for more than 80% of line by 10 kV experiments. Hou et al. [9] utilized the CHB as an
faults in distribution networks [1], [2]. According to the tech- active arc suppression device (AASD). This cascaded structure
nical guidelines for distribution networks [3], the distribution can reduce the harmonic component in the injected current and
networks no longer permit prolonged operation with SLGFs improve the fault current suppression effect [9], [14]. In addi-
and the permanent faults should be isolated quickly. As the tion, the CHB can also compensate for reactive power, offering
capacity of the distribution network increases, the capacitive the advantage of multifunctional integration [15]. Compared
component of the fault current rises [4]. In addition, the high to conventional passive arc suppression devices, the AASDs
are more effective in suppressing fault current. Nevertheless,
Manuscript received 5 February 2024; revised 18 March 2024; when the fault phase detection (FPD) is incorrect, the AASD
accepted 10 April 2024. Date of publication 1 July 2024; date of current
version 30 August 2024. This work was supported in part by the National will raise the nonfault phase voltage, which may result in
Natural Science Foundation of China under Award 51677030. The Associate overcurrent action and insulation damage on the nonfault
Editor coordinating the review process was Dr. Dan Zhang. (Corresponding phase. Consequently, it is crucial to propose an accurate FPD
authors: Jianzhang You; Moufa Guo.) method for active arc suppression technology.
Jiahao Lin and Moufa Guo are with the College of Electrical Engineering
and Automation, Fuzhou University, Fuzhou 350108, China, and also with the Depending on the utilized signals, FPD methods include
Engineering Research Center of Smart Distribution Grid Equipment, Fujian transient-signals-based methods and steady-state-signals-based
Province University, Fuzhou 350108, China (e-mail: [email protected]; methods. Yuan et al. [16] present a wavelet-packet-transform-
[email protected]).
Jianzhang You is with the College of Electrical Engineering and Automa- based FPD method, the optimal frequency band selection
tion, Xiamen University of Technology, Xiamen 361024, China (e-mail: method is proposed to calculate the gray relational degree
[email protected]). of the current decomposition energy of each phase, and
Qiteng Hong is with the Department of Electronic and Electrical
Engineering, University of Strathclyde, G1 1XQ Glasgow, U.K. (e-mail:
the phase with the smallest relational degree is the fault
[email protected]). phase. However, when a high-impedance fault (HIF) occurs,
Digital Object Identifier 10.1109/TIM.2024.3421439 the transient characteristics of SLGFs are not obvious [17],
1557-9662 © 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
3527712 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 73, 2024

which limits the accuracy of the transient-signals-based FPD

method. Therefore, it is common to combine transient and
steady-state signals for FPD [18]. On the other hand, depend-
ing on the characteristics utilized, the FPD methods can be
categorized into amplitude-characteristic-based methods and
angle-characteristic-based methods [18], [19]. In [19], the
relative-angle between each sequence voltage is utilized to
determine the fault phase, and the specific sectors have been
divided to distinguish the fault in different fault scenarios.
Santos et al. [20] perform wavelet transform on voltage and
current decomposing the signal to approximate scale and
detailed scale. In the detail-scale energy spectrum, the energy Fig. 1. Diagram of a simplified distribution network.
of the fault phase is higher than that of the healthy phases,
so the fault phase can be accurately detected. However,
the FPD methods solely based on amplitude characteristics additional signal injection devices, leading to poor economic
or phase angle characteristics pose challenges when applied viability. Moreover, the limited availability of actual data
to varying fault resistances or different distribution network in large quantities makes it difficult to train AI-based FPD
methods to obtain models suitable for different distribution
configurations [21].
networks. Therefore, to reliably determine the fault phase,
Moreover, asymmetrical line parameters can cause neutral
an amplitude-phase-characteristics-based FPD method is pro-
voltage shifts, especially in resonant grounded systems, which
posed in this article. The contributions and advantages of this
may change the fault characteristics of the voltages [22].
FPD method can be summarized as follows.
Therefore, it is necessary to take into account the asymmetry
of line parameters when investigating the methods for FPD. 1) The proposed FPD method incorporates the ampli-
In [23], considering the asymmetry of the line parameters, tude and phase angle characteristics of the three-phase
a novel phase vector analysis method is proposed. The trajecto- source voltage and the zero-sequence voltage, and the
ries of the voltage vectors are analyzed in the case of different amplitude-phase characteristic equation has been derived
grounding methods, and the fault phase can be determined to calculate the phase angle of fault phase voltage. The
by the corresponding FPD criterion. However, the distribution reliability of FPD is not affected by fault resistance,
networks with different grounding methods correspond to dif- asymmetry of line-to-ground parameters, and the load
ferent FPD criteria, which may lead to misjudgment and have current.
2) The phase angle of three-phase source voltage is com-
low applicability. In [24], an active FPD method is proposed,
pared with the phase angle of the fault phase voltage
the fault phase is detected by injecting current and calculating
which is calculated by the amplitude-phase characteristic
the equivalent source voltage of the faulty line. Nevertheless,
equation. The phase with the smallest phase angle dif-
this method requires active construction of FPD conditions
ference is the fault phase, from which the nonthreshold
and an additional signal injection device, which limits the
FPD criterion is constructed.
economy and applicability of this method. Wang et al. [25]
proposed FPD method involving the comparison of phase The proposed FPD method is experimentally validated
voltage magnitude and phase before and after an SLGF. This on 380 V and 10 kV experimental platforms.
method establishes the fault phase selection criteria based The rest of this article is organized as follows. The principle
on different operational scenarios of distribution networks, of the FPD method is introduced in Section II. Section III
and the correct rate of phase selection is not affected by introduces the FPDD and the verification method for esti-
line parameters [25]. However, the practicality of this method mating the accuracy of FPD results. Sections IV and V
needs further confirmation. Fan et al. [26] proposed the FPD are 380 V experimental and 10 kV experimental verification,
method based on the amplitude variation and phase angle respectively. Finally, Section VI provides a conclusion for this
variation of the three-phase voltage, which considers the effect article.
of asymmetry in line parameters, but the FPD criterion is
II. FPD P RINCIPLE
redundant and the correctness of the FPD is low in case of
actual HIFs. A. Neutral Ungrounded System
In addition, because artificial intelligence (AI) algorithms According to Fig. 1, the neutral ungrounded system, reso-
have powerful data processing capabilities, some scholars nant grounded system, and asymmetric line parameters can be
have utilized them in FPD [27], [28]. However, AI-based simulated. Where, U̇ 0 is the zero-sequence voltage. Ė A , Ė B ,
fault diagnosis methods typically require a large amount of and Ė C are the three-phase source voltages. R f is the fault
historical data for learning and training [29], [30], which resistance. L ASC is the inductance of ASC. Ẏ Ai , Ẏ Bi , and Ẏ Ci
presents a significant challenge, as the actual fault data are are the three-phase-to-ground admittance in different feeders,
limited and difficult to collect. respectively, where i = 1, 2, . . . , n. The connection switch
The existing FPD methods mainly suffer from several between ASC and the neutral point is K .
problems, including difficulties in application to different When the connection switch is open, the distribution net-
fault resistances, limited consideration of asymmetric line work shown in Fig. 1 is a neutral ungrounded system.
parameters, and FPD criteria are not consistent under different According to Kirchhoff’s current theorem, the relationship
fault conditions. Particularly, the active FPD methods require between the fault current and phase-to-ground capacitance

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
LIN et al.: AMPLITUDE-PHASE-CHARACTERISTICS-BASED FAULT PHASE DETECTION METHOD 3527712

current can be expressed as In the actual distribution network, the zero-sequence volt-
age and three-phase source voltage can be measured by the
I˙C6 + I˙ B6 + I˙ A6 + I˙ f = 0.


(1)
corresponding transformers. The amplitude and phase angle
According to the node voltage equation, (1) can be rewrit- can be calculated. It can be observed that the calculation
ten as of fault-phase phase angle is not affected by the feeder


1
 loading.
U̇ 0 + Ė C + jωC0C6 Then, the phase-angle difference between − Ė f and − Ė A ,
r0C6
  − Ė B , − Ė C can be calculated, their absolute values are

1
+ U̇ 0 + Ė B + jωC0B6 obtained as
r
 0B6

 D A = abs ̸ (− Ė f ) − ̸ (− Ė A )



 
1
+ (U̇ 0 + Ė A ) + jωC0A6 D B = abs ̸ (− Ė f ) − ̸ (− Ė B ) (11)
r0A6  D = abs ̸ (− Ė ) − ̸ (− Ė )
.


1 C f C
+ U̇ 0 + Ė A =0 (2)
Rf The phase with the minimum phase-angle difference is the
fault phase. Therefore, the FPD criterion for fault phase A, B,
where U̇ 0 + Ė ϕ , ϕ = A, B, and C represent the three-phase and C are as follows, respectively:
voltage, respectively. And each phase-to-ground admittance 
can be defined as 1/r0i6 + jωC0i6 , i = A, B, C. Assuming the  D A = min(D A , D B , DC )

line parameters are symmetrical, the following relationships D B = min(D A , D B , DC ) (12)
can be satisfied:  D = min(D , D , D ).

 C A B C
 Ė A + Ė B + Ė C = 0

C0A6 = C0B6 = C0C6 = C0 (3) B. Resonant Grounding System
0A6 = r 0B6 = r 0C6 = r 0 .

r
When the connection switch K is closed, the distribu-
Then, (2) can be simplified as tion network shown in Fig. 1 is a resonant grounding
  system. Because of the compensation effect of ASC, the
3 1 SLGF characteristics of the zero-sequence voltage and the
U̇ 0 + j3ωC0 + (U̇ 0 + Ė f ) =0 (4)
r0 Rf three-phase source voltage of a resonant grounded system are
different from those of a neutral ungrounded system. Based
where Ė f is the fault phase source voltage and can be
on (5), the zero-sequence voltage of the resonant grounding
rewritten as
system is
− Ė f
U̇ 0 =  . (5) − Ė f
1 + R f r30 + j3ωC0 U̇ 0 = . (13)
1 + R f [3/r0 + j (3ωC0 − 1/ωL ASC )]
In general, the damping rate of the distribution line is less In general, to prevent the occurrence of resonance over-
than 5% [22], namely that, in (5), 3/r0 is much less than 3ωC0 , voltage, the detuning degree of pretuned ASC is typically
and the influence of 3/r0 can be neglected. Therefore, (5) can established at 5% [31]. Therefore, under the normal frequency,
be simplified as the value of (3ωC0 − 1/ωL ASC ) is small, which results in the
U̇ 0 1 phase of − Ė f and U̇ 0 being essentially the same. The phase
= . (6) angle of U̇ 0 can be taken as the phase angle of − Ė f , which
− Ė f 1 + R f j3ωC0
can be expressed as
Therefore, the phase angle of − Ė f can be expressed as ̸ (− Ė f ) = ̸ U̇ 0 . (14)
̸ (− Ė f ) = ̸ U̇ 0 + arctan(R f 3ωC0 ). (7)
Therefore, (14) can be utilized to calculate the phase angle
Meanwhile, the source voltage Ė ϕ can be obtained by of the fault phase source voltage. The fault phase can also be
Ė ϕ = U̇ ϕ − U̇ 0 , ϕ = A, B, C. According to (6), the amplitude determined by the formula (12).
relationship between Ė f and U̇ 0 can be expressed as
U̇ 0 U̇ 0 1 C. Influence of Line Parameters Asymmetry
= =p (8)
− Ė f − Ė φ 1 + (R f 3ωC0 )2 In general, the line parameters of the actual distribution
v network are asymmetric [23], which may lead to neutral
u − Ė φ 2
u
unbalance voltage (U̇ bd ) in the normal operating condition
R f 3ωC0 = t 2
−1 (9) of the distribution network. In this case, when an SLGF
U̇ 0
occurs, the zero-sequence voltage is equal to the superposition
where | Ė φ | is the average amplitude value of the three-phase of the zero-sequence voltage due to the asymmetrical line
source voltage. By substituting (9) into (7), the phase of − Ė f parameters and the zero-sequence voltage generated by SLGF.
can be rewritten as Therefore, the zero-sequence voltage generated by SLGF
v  can be calculated by subtracting the prefault zero-sequence
u − Ė φ 2
u
voltage from the post-fault zero-sequence voltage. The impact
̸ (− Ė f ) = ̸ U̇ 0 + arctant − 1 . (10)
2 of asymmetrical line parameters on the FPD results can be
U̇ 0
mitigated.

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
3527712 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 73, 2024

When accounting for the asymmetry of line parameters in where U̇ f is the fault-phase voltage and Ė f is the fault-phase
a distribution network, (4) can be reformulated as power supply voltage. The formula (19) is rewritten as
   
˙I in = 1 + 3 + j3ωC0 · U̇ f − 3 + j3ωC0 · Ė f .
   
3 3
U̇ 0 + j3ωC0 − U̇ bd + j3ωC0
r0 r0 Rf r0 r0
1 (22)
+ (U̇ 0 + Ė f ) =0 (15)
R
  f  When an SLGF occurs, the fault-phase voltage should be
3 suppressed to zero by AASD to ensure the effectiveness of
U̇ 0 R f + j3ωC0 + 1
r0 fault current suppression.
As for the voltage-based active arc suppression method,
 
3
− U̇ bd R f + j3ωC0 = − Ė f . (16) according to formula (20), the control reference value of U̇ 0 is
r0

By adding −U̇ bd to both sides of the equation, (16) can be U̇ 0_ref = − Ė f . (23)
rewritten as As for the current-based active arc suppression method,

3
 according to formula (21), the control reference of the injected
U̇ 0 − U̇ bd R f ( + j3ωC0 ) + 1 = − Ė f − U̇ bd . (17)


current is
r0  
˙I in_ref = − 3 + j3ωC0 · Ė f . (24)
Generally, U̇ bd is much smaller than Ė f . Then, (17) can be r0
rewritten as
From (22) and (23), it can be obtained that the accurate
− Ė f detection of the fault phase is a prerequisite for the effective
U̇ 0 − U̇ bd ≈ . (18)
1 + R f (3/r0 + j3ωC0 ) suppression of SLGF by the AASD. When the AASD suc-
cessfully suppresses the SLGF, the fault phase voltage will
Therefore, the zero-sequence voltage generated by SLGF be clamped at a lower amplitude. Therefore, it is able to
can be obtained. In a neutral ungrounded system, this verify whether the FPD is correct by the suppression effect
zero-sequence voltage is brought into (10) to obtain the of the AASD.
phase-angle calculated value of the fault phase source voltage,
which can be expressed as
v  B. Design of FPDD
u − Ė φ 2
u
Because the proposed FPD method requires collecting
̸ (− Ė f ) = ̸ (U̇ 0 − U̇ bd ) + arctant − 1 .
2 zero-sequence voltage and three-phase source voltage, the
U̇ 0 − U̇ bd
designed FPDD is installed on the bus cabinet. The FPDD
(19) includes a signal acquisition module and a signal process-
ing module. The signal acquisition module consists of a
In a resonant grounded system, this zero-sequence voltage zero-sequence voltage transformer, three-phase supply voltage
is brought into (14). Consequently, the fault phase can also be transformers, an AD7606 chip, and an STM32F407 microcon-
determined by the formula (12). troller. The signal processing module is the Raspberry Pi.
To realize real-time signal acquisition and calculation, the
III. I MPLEMENTATION OF FPD M ETHOD process of cooperation between the signal acquisition device
and the signal processing device is as follows.
A. Verification and Application of FPD Method The AD7606 chip collects the zero-sequence voltage and
When an SLGF occurs, both AASD and phase-sequence- three-phase power supply voltage and stores them in the
signal-based fault location methods require the correct microcontroller. After storing the data with a length of 0.5 s,
detection of the fault phase. Especially for the AASD, if the it sends the data to the signal processing module by means of
FPD results are inaccurate, the injected current of the AASD direct memory access (DMA) with a baud rate of 1.5 Mbps.
may raise the voltage in the nonfault phase. It may increase the Then, the Raspberry Pi calculates the fault phase by the FPD
amplitude of the nonfault phase current and lead to the relay algorithm and sends the results to the AASD. The time taken
protection system action, which affects the power supply in for data transmission is 0.11 s, and the time taken for fault
the nonfault area. phase calculation is 0.08 s, with the total time consumed less
During the arc suppression process, the compensation cur- than 0.5 s, which can realize real-time signal acquisition and
rent of AASD is injected into the neutral point of the system calculation.
to compensate fault current. According to Fig. 1, the injected The design parameters of FPDD are shown in Table I.
current can be calculated as The connection relationship between FPDD and AASD
  is shown in Fig. 2. The signal acquisition interface in the
˙I in = U̇ f + U̇ 0 3 + j3ωC0 . (20)
designed FPDD is defined according to the distribution net-
Rf r0 work standard. The required voltage signal, which is obtained
from the secondary side of the voltage transformer, is con-
Because U̇ 0 can be expressed as nected to the FPDD through the aviation plug. Meanwhile,
the installation process of FPDD does not require a power
U̇ 0 = U̇ f − Ė f (21) outage and it can be installed within 15 min.

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
LIN et al.: AMPLITUDE-PHASE-CHARACTERISTICS-BASED FAULT PHASE DETECTION METHOD 3527712

FPD algorithm: Calculate the Phase Angle of Fault Phase

Input: Three-phase source voltage ( Ė A , Ė B , Ė C )
Zero-sequence voltage (U̇ 0 )
Neutral unbalance voltage (U̇ bd )
Function definition: Phase angle, Amplitude = STFT(X )
Output: The FPD results
1: for i from 1 to (length ( Ė A )-100) do
2: ̸ Ė A (i), Ė A (i) = STFT( Ė A ( i, i + 100 ))
3: ̸ Ė B (i), Ė B (i) = STFT( Ė B ( i, i + 100 ))
4: ̸ Ė C (i), Ė C (i) = STFT( Ė C ( i, i + 100 ))
5: ̸ U̇ 0 (i), U̇ 0 (i) = STFT(U̇ 0 ( i, i + 100 ))
6: end for
7: if the neutral point is ungrounded then r
|− Ė φ |2
8: ̸ (− Ė f ) = ̸ (U̇ 0 − U̇ bd ) + arctan( − 1) (19)
|U̇ 0 −U̇ bd |2
9: end if
10: if the neutral point is grounded by ASC then
11: ̸ (− Ė f ) = ̸ (U̇ 0 ) (14)
12: end if

13: D A = abs ̸ (− Ė f ) − ̸ (− Ė A )


Fig. 2. Structural diagram of relevant experimental equipment.

14: D B = abs ̸ (− Ė f ) − ̸ (− Ė B )

15: DC = abs ̸ (− Ė f ) − ̸ (− Ė C )


16: Dref = min( DA , DB , DC )
17: if DA = Dref then
18: fault phase = “Phase A”
19: else if DB = Dref then
20: fault phase = “Phase B”
21: else
22: fault phase = “Phase B”
23: end if
24: return

TABLE I
D ESIGN PARAMETERS OF FPDD

Fig. 3. Principle of FPD method and FPD accuracy verification method.

which is compared with the measured value of the three-phase

source voltage, respectively.
Step 3: The fault phase can be determined by the defined
FPD criterion. Then, the FPD result is sent to AASD to verify
C. FPD Process the accuracy of the FPD result.
Step 4: The corresponding control target is utilized accord-
The proposed FPD algorithm is utilized to identify the ing to the control method of AASD. And the compensation
fault phase when an SLGF occurs. Where, the short-time current is injected.
Fourier transform (STFT) is utilized to calculate the phase Step 5: If the fault phase voltage is effectively suppressed
angle and amplitude of signal. Then, the AASD determines at low amplitude, the FPD result is correct.
the control target according to the FPD results. It can be seen
that both the voltage-based active arc suppression method and IV. V ERIFICATION OF P HYSICAL
the current-based active arc suppression method require the E XPERIMENTAL P LATFORM
accurate FPD. The schematic of the proposed FPD method
and its result validation is shown in Fig. 3. A. Physical Experimental Platform
The process of FPD and the verification of FPD result is To validate the effectiveness of the proposed FPD detec-
shown in Fig. 4. And the detailed steps of the proposed FPD tion method, the physical experimental platform is designed
method are as follows: based on the principle of similarity. The rated voltage of the
Step 1: The zero-sequence voltage and three-phase source experimental platform is 380 V, and the parameters of the com-
voltage are collected in real time and the SLGF is detected. ponents utilized in this experimental platform are calculated
Step 2: When an SLGF occurs, the phase angle of fault based on the actual distribution network. It provides a viable
phase source voltage is calculated according to (10) or (14), experimental plat for investigating the distribution network’s

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
3527712 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 73, 2024

Fig. 6. Structure of the physical experimental platform.

Fig. 4. Process of FPD and FPD results verification.

Fig. 7. FPD results. (a) R f = 10 k, asymmetry degree: 3.74%.

Fig. 5. Diagram of a simulated distribution network. (b) R f = 10 , asymmetry degree: 3.74%. (c) R f = 10 k, asymmetry
degree: 8.3%. (d) R f = 10 , asymmetry degree: 8.3%. (e) R f = 10 k,
asymmetry degree: 14%. (f) R f = 10 , asymmetry degree: 14%.
operation principles and fault characteristics simulation. The
experimental platform includes six feeders and the feeders
power supply voltage is compared with the measured value
include both resistance and inductive loads, which are utilized
of the phase angle of the three-phase source voltage. Based
to simulate the 10 kV distribution network. The diagram of
on the FPD criterion, the phase with the smallest difference
the simulated distribution network is illustrated in Fig. 5.
between the phase-angle measured value and the phase-angle
In addition, the experimental platform is equipped with a
calculated value of the fault phase source voltage is the fault
fault-generating device, which can simulate the SLGF with
phase. Under different fault conditions, the calculated phase
different resistances and the SLGF with arc. The structure of
angle of the fault phase source voltage and the measured phase
the experimental platform is shown in Fig. 6.
angle of the three-phase source voltage are depicted in Fig. 7.
It demonstrates that the proposed FPD method can accurately
B. Neutral Ungrounded System
identify the fault phase in the case of different fault resistances
According to the experimental platform depicted in Fig. 6, and line parameters with different degrees of asymmetry.
the accuracy of the proposed FPD method can be validated
under different fault conditions, such as different line param- C. Resonant Grounding System
eter asymmetries (3.74%, 8.3%, and 14%) and different fault
resistances (10 k and 10 ). The three-phase source voltage According to (20) and Fig. 1, the following relationships
and zero-sequence voltage are collected by FPDD depicted can be satisfied:
(
in Fig. 2. In addition, according to the proposed FPD method, U̇ f = I˙ f · R f
the phase angle of the fault phase source voltage can be (25)
I˙ f = ( I˙ f _0 − I˙ L_ASC ) = U̇ 0 · ωCP − 1 ωL ASC


calculated. The phase-angle calculated value of the fault phase

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
LIN et al.: AMPLITUDE-PHASE-CHARACTERISTICS-BASED FAULT PHASE DETECTION METHOD 3527712

TABLE II
FPD R ESULTS U NDER D IFFERENT S CENARIOS

Fig. 9. FPD results of traditional methods in a neutral ungrounded system.

Fig. 8. FPD results. (a) Asymmetry degree: 3.74%, fault phase: phase A. Voltage amplitude reduction in (a) one phase and (b) two phases.
(b) Asymmetry degree: 3.74%, fault phase: phase B. (c) Asymmetry
degree: 3.74%, fault phase: phase C. (d) Asymmetry degree: 7.8%, fault phase:
phase A.
In the fault phase, the means and variances of Dx , x = A,
B, C are the smallest, which indicates that the calculation
where I˙ f _0 represents the fault current before compensation, results, Dx , x = A, B, C, of fault phase and normal phases are
CP is the total capacitance to ground, ωCP = 3ωC0 , and well distinguishable, ensuring the accuracy of FPD. It can be
ω is the dominant frequency of distribution network. The obtained from Tables II and III that the fault phase is detected
expression for zero-sequence voltage can be reorganized as with high accuracy and robustness when the proposed FPD
method is utilized.
Ė f
U̇ 0 =
. (26)
R f · 1 + 1 ωL ASC − ωCP

D. Comparison With Other Phase Detection Methods
After an SLGF occurs, the initial dominant frequency is In traditional FPD methods, the amplitude of the phase
high, but the frequency of the system exhibits decaying char- voltage is utilized as an important criterion. In the neutral
acteristics. When ω = 50 Hz, the total capacitive reactance of ungrounded system, the voltage amplitude of the fault phase
the system is approximately equal to the inductive reactance of decreases. However, if two phases decrease in voltage ampli-
ASC, namely 1/ωCP ≈ ωL ASC . Therefore, it can be obtained tude and one phase increases in voltage amplitude, the phase
that the phase angle of the zero-sequence voltage is close to lagging the phase with the increased voltage amplitude is the
the phase angle of the fault phase source voltage only when fault phase. The schematic of fault phase selection for the
the capacitive component of the fault current is compensated. traditional method in a neutral ungrounded system is shown
According to [32], typically three cycles after SLGFs, the in Fig. 9.
capacitive component of fault current can be compensated. In the figure, the U̇ ϕ(1) , U̇ ϕ(2) , and U̇ ϕ(3) indicate the phase
In resonant grounding, it is accurate to employ the phase angle voltage. In resonant grounded systems, the leading phase
of the zero-sequence voltage to represent the calculated value of the phase exhibiting a rising voltage amplitude can be
of the fault phase source voltage after several cycles of SLGF defined as the fault phase [21]. However, this method has
occurrence. Therefore, the phase-angle calculated value of the inconsistent criteria under different conditions, which leads
fault phase source voltage and the phase-angle measured value to its low applicability. Meanwhile, when the HIFs occur, the
of the three-phase source voltage are depicted in Fig. 8. amplitude characteristics of phase voltage are relatively low,
To better represent the change in phase angle of zero- which may lead to misjudgment of traditional FPD methods.
sequence voltage, the ASC is connected six cycles after an The waveform characteristics of phase voltage under HIFs are
SLGF occurs. It can be obtained that the proposed FPD illustrated in Fig. 10.
method can also be applied to resonant grounding systems, In [16], a wavelet-packet-transform-based FPD method is
relevant experimental conditions and FPD results are listed proposed and the gray relational degree of each phase current
in Table II. The fault resistance values are selected according in the optimal frequency band is calculated and utilized as the
to the criteria of [33]. FPD criterion. However, the current-based method is suscepti-
The results of FPD validate the effectiveness of the proposed ble to interfering signals and the FPD correctness is low in the
method. In addition, the reliability of the proposed FPD case of HIFs. Guo et al. [24] propose an active FPD method
method can be verified by the means and variances of Dx , based on a CHB device, which can actively construct the FPD
x = A, B, C. In these experimental cases, the means for each criterion. However, this method requires an additional signal
phase are denoted as D A_m , D B_m , and DC_m , respectively. injection device with low applicability [24]. Wang et al. [25]
And the variances are denoted as D A_v , D B_v , and DC_v , analyzed the change characteristics of each phase voltage pre
respectively. The calculation results are presented in Table III. and post SLGF and determined the fault phase based on

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
3527712 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 73, 2024

TABLE III
M EANS AND VARIANCES OF D A , D B , AND DC

TABLE IV
FPD R ESULTS OF T HESE FPD M ETHOD

TABLE V
C ORRECT FPD R ATE OF T HESE M ETHODS

Taking the asymmetry of the line parameters as 3.74%

as an example, the FPD results of these methods are listed
in Table IV. In addition, these methods are utilized to test
the experimental data in Table II, and the correct FPD rate
for each method is shown in Table V. It can be obtained that
although the methods proposed in [24] and [25] have good
fault phase selection performance, the method proposed in this
article possesses the highest FPD correctness. Meanwhile, the
FPD method in [24] requires an active signal injection device,
which limits the generalizability of this method.
Fig. 10. Phase voltage waveform when transition resistance is 3000 .
(a) Neutral ungrounded system. (b) Resonant grounding system.
V. V ERIFICATION OF 10 K V E XPERIMENTAL P LATFORM
A. Experimental Platform
the trend of their amplitude and phase angle. However, the To further validate the accuracy of the proposed method
applicability of this method in actual distribution networks for detecting the fault phase. The experimental cases are
needs further verification. In [26], an FPD method based on conducted on a 10 kV experimental platform under various
amplitude variation analysis and phase angle variation analysis fault conditions. The experimental platform simulates a 10 kV
is proposed, but the FPD criterion of this method is more distribution network, which includes four feeders that are
complicated and the correctness of FPD is not as good as that simulated by the pi-type equivalent model. The SLGFs such
of the former method in the case of HIFs. as arc faults, metallic faults, and resistance faults can be

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
LIN et al.: AMPLITUDE-PHASE-CHARACTERISTICS-BASED FAULT PHASE DETECTION METHOD 3527712

TABLE VI
FPD R ESULTS IN THE N EUTRAL U NGROUNDED S YSTEM

TABLE VII
FPD R ESULTS IN THE R ESONANT G ROUNDING S YSTEM

Fig. 11. 10 kV experimental platform. (a) Main wiring diagram. (b) Physical
image.

TABLE VIII
FPD R ESULTS FOR D IFFERENT A RC S CENARIOS

Fig. 12. Prototype of 10 kV active suppression device. (a) Control schematic.

(b) Physical diagram.

simulated by this experimental platform. The main wiring

diagram and physical diagram of the 10 kV experimental
platform are illustrated in Fig. 11. In Fig. 11(a), Tvr is the
voltage regulating transformer, Tyd is the step-up transformer,
Tzt is the grounding transformer, L 1x , L 2x , L 3x , and L 4x
are the line inductance, R1 , R2 , R3 , and R4 are the ground
leakage resistance, and C1x , C2x , C3x , and C4x are the ground the difference between the calculated phase angle and the
capacitances, where x = 1, 2, 3. measured phase angle can be obtained by formula (11). The
In addition, a prototype of the 10 kV AASD is con- calculated results and related experimental waveforms for FPD
structed to verify the accuracy of the FPD results. The AASD are shown in Fig. 13.
consists of the CHB, CHB controllers, and other associated After fault arc suppression, the fault phase voltage can be
industrial-grade components and a real-time control system. diminished to a significantly low level, and the correctness of
The communication between the controller and the H-bridge the FPD results can be validated. The FPD results for various
module is realized through optical fiber. Fig. 12 depicts the fault conditions are listed in Table VI.
control schematic and physical diagram of the 10 kV AASD
prototype. C. Resonant Grounding System
In the resonant grounding system, the phase angle of
B. Neutral Ungrounded System the fault phase source voltage can also be calculated by
When an SLGF occurs, the calculated phase angle of formula (12). To validate the accuracy of the FPD results,
fault source voltage can be obtained by formula (10). Then, the fault phases A and B are considered, along with fault

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
3527712 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 73, 2024

Fig. 13. Experimental waveform and the FPD calculated results in a neutral Fig. 14. Experimental waveform and the FPD calculated results in the
ungrounded system. (a) Grounding resistance is 500 , the fault phase is resonant grounding system. (a) Grounding resistance is 500 , fault phase
phase A. (b) Grounding resistance is 10 k, the fault phase is phase B. is phase A. (b) Grounding resistance is 10 k, fault phase is phase B.

TABLE IX
SLGF PARAMETERS OF D IFFERENT S CENARIOS

Fig. 15. Different arc experimental scenarios and their arc combustion
processes. (a) Branch arc. (b) Stone arc. (c) Sand arc.
resistances of 500  and 10 k. Fig. 14 presents the FPD
results and corresponding experimental waveforms. In the
actual distribution network, the ASC is immediately connected
detection of the fault phase. Therefore, different arc fault
to the neutral. Therefore, during the 10 kV experiment, the
scenarios are simulated by the system shown in Fig. 11,
ASC is connected at the moment when the SLGF occurs.
including branch arc, stone arc, and sand arc, to verify the
According to Fig. 14, the fault phase voltage is effectively
effectiveness of the proposed FPD method. The simulated
suppressed, which verifies the applicability of the proposed
scenarios are shown in Fig. 15.
FPD method in resonant grounding systems. Meanwhile, the
And taking an SLGF occurs in phase A as an example,
FPD results for various fault conditions are listed in Table VII.
the FPD results are shown in Table VIII. The different fault
scenarios have been simulated by the 10 kV experimental
D. Arc Experimental Cases in Different Scenarios platform. In the actual distribution network, the asymmetry
In an actual distribution network, if an SLGF occurs, it may of the line parameters is generally no more than 5% [23], and
often be accompanied by arcing. In this situation, the voltage the high line parameters asymmetry may trigger a large neutral
and current contain high harmonic components, which brings displacement voltage and damage the experimental equip-
great challenges to the measurement of the signal and the ment when simulating an SLGF. Therefore, in the simulated

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
LIN et al.: AMPLITUDE-PHASE-CHARACTERISTICS-BASED FAULT PHASE DETECTION METHOD 3527712

TABLE X
M EANS AND VARIANCES OF D A , D B , DC FOR 10 kV E XPERIMENTAL DATA

TABLE XI
C ORRECT R ATE OF FPD M ETHODS FOR 10 kV E XPERIMENTAL DATA

experimental cases, the asymmetry of line parameters is The correct detection of the fault phase for SLGF holds
assumed to be 3.74% and 5.0%, respectively. The simulated great importance in preserving the steady operation of distri-
experimental cases are described in Table IX. bution networks. The compensation current of ASC cannot be
To further validate the robustness and reliability of the changed abruptly, and the phase angle of the zero-sequence
proposed FPD method, the means and variances of D A , D B , voltage is still changing during the transient process. There-
and DC are calculated separately, and the calculation results fore, the proposed FPD method takes a relatively long time
are listed in Table X. It can be seen from Tables IX and X to accurately select the fault phase in resonant grounding
that the proposed FPD method still has excellent performance systems. Therefore, in future research, the phase selection
in the 10 kV experimental cases. speed of the FPD method in resonant grounding systems will
be further improved.
E. Comparison Analysis
R EFERENCES
The FPD results of the proposed method and the compared [1] J. Lin, M. Guo, and Z. Zheng, “Active location method for single-line-
methods are listed in Table XI. The proposed method is to-ground fault of flexible grounding distribution networks,” IEEE Trans.
unaffected by asymmetric line-to-ground parameters, ground Instrum. Meas., vol. 72, 2023, Art. no. 3524712.
fault resistance, and load current, which demonstrates high [2] Y.-L. Liang, K.-J. Li, Z. Ma, and W.-J. Lee, “Typical fault cause recog-
nition of single-phase-to-ground fault for overhead lines in nonsolidly
reliability in neutral ungrounded systems as well as resonant earthed distribution networks,” IEEE Trans. Ind. Appl., vol. 56, no. 6,
grounding systems. Meanwhile, it can also be obtained from pp. 6298–6306, Nov. 2020.
Table XI that the FPD method proposed in this article still has [3] Technical Guidelines for Distribution Network, Standard Q/GDW 10370-
the highest FPD correct rate in the 10 kV experimental data, 2016, State Grid Corporation of China, Beijing, China, 2016.
[4] J. Lin, M. Guo, Q. Hong, and R. Jiang, “An earth fault diagnosis method
which indicates the good applicability of this method. based on online dynamically calculated thresholds for resonant ground
systems,” IEEE Trans. Smart Grid, vol. 15, no. 4, pp. 3459–3473,
VI. C ONCLUSION Jul. 2024, doi: 10.1109/TSG.2023.3346453.
[5] Q. Guo et al., “An overview of series-connected power electronic
Considering the line parameter imbalance, this article anal- converter with function extension strategies in the context of high-
yses the magnitude and phase angle characteristics of the penetration of power electronics and renewables,” Renew. Sustain.
zero-sequence voltage and the three-phase source voltage Energy Rev., vol. 156, Mar. 2022, Art. no. 111934.
when SLGF occurs. To accurately determine the fault phase, [6] A. Nikander, “Development and testing of new equipment for faulty
phase earthing by applying RTDS,” IEEE Trans. Power Del., vol. 32,
an amplitude-phase-characteristics-based FPD method is pro- no. 3, pp. 1295–1302, Jun. 2017.
posed. After theoretical analysis and experimental verification, [7] M. Rošer and G. Štumberger, “Improving the power supply reliability
the following conclusions can be obtained. The proposed in resonant earthed systems by fault current path control established
FPD method integrates the amplitude and phase angle char- through faulted phase earthing switch,” Int. J. Elect. Power Energy Syst.,
vol. 64, pp. 714–722, Jan. 2015.
acteristics of zero-sequence voltage and three-phase source [8] K. Winter, “The RCC ground fault neutralizer—A novel scheme for fast
voltage to formulate a nonthreshold phase selection criterion, earth-fault protection,” in Proc. 18th Int. Conf. Exhib. Electr. Distrib.
which provides high accuracy for FPD under line parameters (CIRED), Turin, Italy, Jun. 2005, pp. 6–9.
imbalance. And because the source-side voltage is utilized, the [9] Y. Hou et al., “Adaptive active voltage-type arc suppression strategy con-
sidering the influence of line parameters in active distribution network,”
FPD results are unaffected by the fault resistance and the load IEEE Trans. Ind. Electron., vol. 70, no. 5, pp. 4799–4808, May 2023.
current. The experimental data verify the effectiveness of the [10] W. Wang, X. Zeng, L. Yan, X. Xu, and J. M. Guerrero, “Principle and
proposed method. In engineering applications, the three-phase control design of active ground-fault arc suppression device for full
source voltage and the zero-sequence voltage have been mea- compensation of ground current,” IEEE Trans. Ind. Electron., vol. 64,
no. 6, pp. 4561–4570, Jun. 2017.
sured in substations, which does not require additional signal [11] K. Chen and X. Xiao, “Variable impedance fault current bypass circuit
measurement equipment, making the proposed FPD method for fault arc suppression in distribution network,” IET Gener., Transmiss.
and the designed FPDD highly feasible. Distrib., vol. 13, no. 18, pp. 4214–4221, Sep. 2019.

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.
3527712 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 73, 2024

[12] J. Rorabaugh et al., “Testing the increased sensitivity and energy [32] X. Wei, X. Wang, J. Gao, D. Yang, K. Wei, and L. Guo, “Faulty feeder
reduction of a ground fault neutralizer for wildfire mitigation,” in Proc. detection for single-phase-to-ground fault in distribution networks based
IEEE Power Energy Soc. Gen. Meeting (PESGM), Denver, CO, USA, on transient energy and cosine similarity,” IEEE Trans. Power Del.,
Jul. 2022, pp. 1–6. vol. 37, no. 5, pp. 3968–3979, Oct. 2022.
[13] B. Fan et al., “Principle and control design of a novel hybrid arc [33] Regulatory Impact Statement: Bushfire Mitigation Regulations Ammend-
suppression device in distribution networks,” IEEE Trans. Ind. Electron., ment, Dept. Econ. Develop., Jobs, Transp. Resour., Acil Allen
vol. 69, no. 1, pp. 41–51, Jan. 2022. Consulting, Australia, Nov. 2015.
[14] B.-L. Zhang, M.-F. Guo, Z.-Y. Zheng, and Q. Hong, “Fault current
limitation with energy recovery based on power electronics in hybrid
AC–DC active distribution networks,” IEEE Trans. Power Electron.,
vol. 38, no. 10, pp. 12593–12606, Oct. 2023.
[15] J. Li, J. Chen, and C. Gong, “An optimized reactive power compensation Jiahao Lin (Student Member, IEEE) was born in
strategy to extend the working range of CHB multilevel grid-tied Fujian, China, in 1997. He received the B.S. degree
inverters,” IEEE Trans. Power Electron., vol. 38, no. 4, pp. 5500–5512, in electric machines and electric apparatus from
Apr. 2023. Fuzhou University, Fuzhou, Fujian, in 2019, where
[16] W. Yuan, Y. Li, L. Xu, T. Li, and X. Chen, “A fast faulty phase selection he is currently pursuing the Ph.D. degree in electric
method considering fault tolerance for single phase to ground fault machines and electric apparatus.
in distribution networks,” IEEE Trans. Instrum. Meas., vol. 72, 2023, His main research interests include the detection
Art. no. 3533212. and location of single-line-to-ground faults in distri-
[17] J. Li, Y. Liu, C. Li, D. Zeng, H. Li, and G. Wang, “An FTU-based bution networks.
method for locating single-phase high-impedance faults using transient
zero-sequence admittance in resonant grounding systems,” IEEE Trans.
Power Del., vol. 37, no. 2, pp. 913–922, Apr. 2022.
[18] K. Xu, Z. Zhang, Q. Lai, and X. Yin, “Fault phase selection method
applied to tie line of renewable energy power stations,” IET Gener.,
Transmiss. Distrib., vol. 14, no. 13, pp. 2549–2557, Mar. 2020. Jianzhang You (Member, IEEE) was born in
[19] S. Huang, L. Luo, and K. Cao, “A novel method of ground fault phase Fujian, China, in 1990. He received the M.S. and
selection in weak-infeed side,” IEEE Trans. Power Del., vol. 29, no. 5, Ph.D. degrees in electrical engineering from Fuzhou
pp. 2215–2222, Oct. 2014. University, Fuzhou, China, in 2016 and 2023,
[20] W. C. Santos, F. V. Lopes, N. S. D. Brito, and B. D. Souza, “High- respectively.
impedance fault identification on distribution networks,” IEEE Trans. Since 2023, he has been with Xiamen Univer-
Power Del., vol. 32, no. 1, pp. 23–32, Feb. 2017. sity of Technology, Xiamen, China, where he is
[21] L. Zhang et al., “Fault phase identification of non-solidly grounding currently a Lecturer with the College of Electrical
system considering system asymmetry,” Electr. Power Automat. Equip., Engineering and Automation. His research interests
vol. 39, no. 4, pp. 24–29, Apr. 2019. include information processing, protection control,
[22] A. Gargoom, A. M. T. Oo, and M. Cavanagh, “A method for calculating and flexible arc suppression of single-line-to-ground
the asymmetry in the shunt parameters of power lines in compen- faults in distribution networks.
sated distribution networks,” IEEE Trans. Power Del., vol. 35, no. 5,
pp. 2168–2176, Oct. 2020.
[23] B. Xu et al., “Grounding phase determination in non-effective grounding
systems in case of unsymmetrical voltage,” Trans. China Electrotech.
Soc., vol. 26, no. 12, pp. 175–182, Dec. 2011. Moufa Guo (Member, IEEE) was born in Fujian,
[24] M.-F. Guo, W.-Q. Cai, Z.-Y. Zheng, and H. Wang, “Fault phase selection China, in 1973. He received the B.S. and
method based on single-phase flexible arc suppression device for asym- M.S. degrees in electrical engineering from Fuzhou
metric distribution networks,” IEEE Trans. Power Del., vol. 37, no. 6, University, Fuzhou, China, in 1996 and 1999,
pp. 4548–4558, Dec. 2022. respectively, and the Ph.D. degree in electrical engi-
[25] W. Wang, X. Gao, B. Fan, X. Zeng, and G. Yao, “Faulty phase detection neering from Yuan Ze University, Taoyuan, Taiwan,
method under single-line-to-ground fault considering distributed param- in 2018.
eters asymmetry and line impedance in distribution networks,” IEEE Since 2000, he has been with Fuzhou University,
Trans. Power Del., vol. 37, no. 3, pp. 1513–1522, Jun. 2022. where he is currently a Professor with the College of
[26] B. Fan et al., “Faulty phase recognition method based on phase-to- Electrical Engineering and Automation. His research
ground voltages variation for neutral ungrounded distribution networks,” interests include information processing, protection
Electr. Power Syst. Res., vol. 190, Jan. 2021, Art. no. 106848. control, and flexible arc suppression of single-line-to-ground faults in distri-
[27] Z. Q. Bo, R. K. Aggarwal, A. T. Johns, H. Y. Li, and Y. H. Song, bution networks.
“A new approach to phase selection using fault generated high frequency
noise and neural networks,” IEEE Trans. Power Del., vol. 12, no. 1,
pp. 106–115, Jan. 1997.
[28] K. Lout and R. K. Aggarwal, “Current transients based phase selection
and fault location in active distribution networks with spurs using Qiteng Hong (Senior Member, IEEE) received the
artificial intelligence,” in Proc. IEEE Power Energy Soc. Gen. Meeting, B.Eng. and Ph.D. degrees (Hons.) in electronic
Vancouver, BC, Canada, Jul. 2013, pp. 1–5. and electrical engineering from the University of
[29] M.-F. Guo, J.-H. Gao, X. Shao, and D.-Y. Chen, “Location of single-line- Strathclyde, Glasgow, U.K., in 2011 and 2015,
to-ground fault using 1-D convolutional neural network and waveform respectively.
concatenation in resonant grounding distribution systems,” IEEE Trans. He is currently a Senior Lecturer (Associate Pro-
Instrum. Meas., vol. 70, 2021, Art. no. 3501009. fessor) at the University of Strathclyde. His main
[30] R. Jiang, Y. Wang, X. Gao, G. Bao, Q. Hong, and C. D. Booth, research interests include power system protection
“AC series arc fault detection based on RLC arc model and convolutional and control in future networks with high penetration
neural network,” IEEE Sensors J., vol. 23, no. 13, pp. 14618–14627, of renewables.
Jul. 2023. Dr. Hong was a Regular Member of the completed
[31] Q. Li et al., “Incipient fault detection in power distribution system: A CIGRE WG B5.50. He is a member of IEEE Working Group P2004 and the
time-frequency embedded deep-learning-based approach,” IEEE Trans. IEEE Task Force on Cloud-Based Control and Co-Simulation of Multi-Party
Instrum. Meas., vol. 72, 2023, Art. no. 2507914. Resources in Energy Internet.

Authorized licensed use limited to: Makerere University Library. Downloaded on October 25,2024 at 16:14:57 UTC from IEEE Xplore. Restrictions apply.