10 1109@jphotov 2019 2892189

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 9, NO.

2, MARCH 2019 513

A Comparative Evaluation of Advanced Fault


Detection Approaches for PV Systems
Dhanup S. Pillai , Frede Blaabjerg , Fellow, IEEE, and Natarajan Rajasekar , Member, IEEE

Abstract—Even with the consistent growth in global photovoltaic point tracking (MPPT) [2], array reconfiguration (AR) [3], [4],
(PV) capacity, the necessity for fault detection in PV systems has and grid interconnection [5], [6] of PV systems. Usually, for
not been widely addressed regardless of its importance. With Inter- dc side protection, PV installations follow standards, such as
national Electrotechnical Commission, Institute of Electrical and
Electronics Engineers, and National Electric Code protection stan- National Electric Code (NEC) 690, IEEE standard 1374, and
dard recommendations for PV systems being vulnerable to fault European International Electrotechnical Commission Standard
occurrences, advanced detection techniques are inevitable in PV (IEC) 62548 [7]–[9]. Yet, faults in PV arrays are unavoidable
systems to avail guaranteed protection from electric shocks and and provoked recent fire events due to long term undetected
fire hazards. Even though numerous fault detection techniques faults [10], [11]. In this paper, the fault detection in PV systems
have been conceptualized in last few years, these techniques are
yet to be classified and quantified in a common platform. Hence, receives colossal importance in endorsing the future growth of
this paper takes up an initiative to study, classify and analyze the PV installations.
advanced fault detection approaches available in literature. Each Presently, PV protection standards available intend to protect
fault detection technique is segregated based on the detection ap- PV arrays from three major faults line-line (LL), line-ground
proach and are reviewed with respect to: types of faults detected; (LG), and arc faults. Even then, PV arrays are susceptible to nu-
detection time; sensor requirement; procedural complexity; de-
tection variables; and level of protection achieved. Furthermore, a merous other fault possibilities open circuit (OC) faults, bypass
compatibility study is conducted to evaluate the effectiveness of ad- diode failures, shade faults, hotspots, connection failures, and
vanced fault detection techniques for protection against line-line, degradation faults. Therefore, in addition to existing methods,
line-ground, and arc faults that are most common in PV systems. several advanced fault detection techniques have been proposed
Overall, this investigation serves as a valuable reference for re- in recent times to avail reliable protection. Further, each tech-
searchers to improve fault detection possibilities in PV systems.
nique propounded for fault detection possesses conceptual dif-
Index Terms—Fault diagnosis, photovoltaic power systems, ferences on the basic approach that they use to detect faults. Un-
protection and solar power generations, protection standards.
fortunately in the literature, these approaches are yet to be classi-
fied and quantified to analyze its effectiveness. Hence, a detailed
study becomes inevitable to classify, analyze, and evaluate the
I. INTRODUCTION effectiveness of advanced fault detection approaches. Therefore,
HOTOVOLTAIC (PV) systems contribute 2% of the this paper attempts to provide an in-depth analysis on various
P world’s total energy consumption, i.e., 303 GW of the
power generation capacity. Being the foremost among existing
advanced fault detection approaches and conducts a detailed
performance evaluation as well as a comparative study with re-
renewable energy resources, power reduction due to fault occur- spect to the conventional protection schemes. Furthermore, the
rences have been a major concern to the PV system efficiency detection techniques are categorized based on their detection
and its reliability [1]. Unfortunately, awareness on the PV fault approach and reliability of each technique is reviewed based on-
detection is relatively very slow compared to the rapid advance- types of faults detected, detection time, sensor requirements, ap-
ments that have been made so far in the field of maximum power plicability, detection variables, and level of protection. To com-
prehend, a compatibility study is also carried out to analyze the
effectiveness of advanced fault detection approaches in detect-
ing LL, LG, and arc faults that are most common in PV systems.
Manuscript received November 1, 2018; revised December 12, 2018; ac-
cepted January 7, 2019. Date of publication January 24, 2019; date of current II. TYPICAL PV ARRAY FAULTS, PROTECTION SCHEMES, AND
version February 18, 2019. The work was supported by the Vellore Institute of
Technology, Vellore, through “VIT SEED GRANT.” (Corresponding author:
NECESSITY OF ADVANCED DETECTION METHODS
N. Rajasekar.) A. PV Array Faults
D. S. Pillai and N. Rajasekar are with the Solar Energy Research Cell,
Department of Energy and Power Electronics, School of Electrical Engineer- Photovoltaic systems are frequently challenged by electrical
ing, Vellore Institute of Technology, Vellore 632014, India (e-mail:, dhanup.
research@gmail.com; nrajasekar@vit.ac.in). fault occurrences that arise due to abnormalities in the internal
F. Blaabjerg is with the Department of Energy Technology, Aalborg Univer- system configuration. Among many, the most common faults
sity, Aalborg 9100, Denmark (e-mail:,fbl@et.aau.dk). that occur in PV array are LG, LL, OC, and arc fault. LG
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. faults occur due to accidental connection between equipment
Digital Object Identifier 10.1109/JPHOTOV.2019.2892189 grounding conductor and current carrying conductor [12]–[14].
2156-3381 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
514 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 9, NO. 2, MARCH 2019

Fig. 1. Various faults occurring in a PV Array.

While, LL faults are created by accidental short-circuit be- B. Conventional Protection Schemes and Its Limitations
tween any two points in the PV array; and it can be either
intra string or cross string faults. Moreover, any open-circuit Various conventional protection schemes available to protect
between two panels induces OC faults and interrupts the current PV arrays against fault occurrences can be typically classified
flow [15]. Usually, OC faults occur as an after effect of short- into three as follows:
circuit faults or due to a connection failure in the PV array. 1) use of standard protection devices;
Arc faults can be either of series or parallel-type created due 2) real-time/offline testing using diagnostic tools;
to open-circuit faults in a string or due to insulation breakdown 3) experimental analysis to detect the anomalies presents in
between two points of adjacent strings [16]–[18]. the system parameters.
Unlike electrical faults, shade faults are created by the un- According to NEC article 690, over current protection de-
even distribution of shades over the PV array. Partial shading vice (OCPD), ground fault detection and interruption (GFDI)
(PS) occurs due to any one of the following events [19], such as fuse and arc fault circuit interrupter (AFCI) are usually installed
building shadows, tree shadows, passage of clouds, bird drop- in PV arrays for protection against LL, LG, and arc faults, re-
pings, and dust accumulation on the panel surface. In most spectively. However, extensive studies carried out in [1], [10],
cases, PS is temporary; however, prolonged shade occurrences and [11] revealed that the inimitable operating characteristics
due to dust accumulation and bird droppings lead to forma- of PV systems frequently deceive these protection devices. For
tion of hot spots and degrade PV panels if remain undetected. instance, OCPDs fail to detect LL faults when I-V mismatch
Besides, low power rating of bypass diodes induces frequent level is less than 40%, faults occur in low irradiation levels, and
failures that constitute to bypass diode faults [11]. In addition, PV systems are installed with blocking diodes. Similarly, GFDI
any improper switching of the MPPT controller and age factor fuses are ineffective at low irradiance and high fault impedance
increases the thermal stress on switch that leads to converter levels. Moreover, the MPPT optimization limits the capability
faults [20]. Also, any abnormal charging conditions or internal of both OCPDs and GFDI fuses [1], [10]. On the other hand,
cell damage create battery bank fault. Even though battery bank test reports on the confines of arc fault detection in PV sys-
faults do not affect the PV array, it adversely affects the energy tems using AFCIs are comprehensively studied and detailed in
management system. For clarity, all possible electrical faults that [21]. Distinctive from the above, offline testing using insulation
occur in the dc side of a PV system are pictorially represented measurement device and online testing using circuit tracers are
in Fig. 1. also employed for ground fault and open-circuit fault detection,
PILLAI et al.: COMPARATIVE EVALUATION OF ADVANCED FAULT DETECTION APPROACHES FOR PV SYSTEMS 515

TABLE I
DIFFERENTIATION OF ADVANCED DETECTION TECHNIQUES

respectively [10]. However, despite of its advantages, these


methods also suffer from some prominent disadvantages, such as
large detection time, manual intervention necessity, and power
flow obstruction. On the whole, neither of the conventional tech-
niques available till date is trustworthy in protecting PV systems.
Therefore, reliable fault detection techniques/algorithms having
the ability to detect multiple faults without interfering with the
power generation is a mandatory requirement for reliability and
safety of PV systems.
Fig. 2. Flowchart for model-based difference measurement approach.
III. ADVANCED FAULT DETECTION APPROACHES
Motivated by the protection concerns, several new fault de- In (1), “k” is the tolerance factor incorporated in the detection
tection techniques have been proposed for PV systems recently. rule to avoid nuisance tripping. For a better understanding, the
Each advanced technique propounded for fault detection in PV flowchart of the MBDM approach is given in Fig. 2. Further,
systems possesses conceptual differences on their basic ap- following the MBDM approach and using different detection
proach that they use to detect faults. Hence, this paper classi- variables, various fault detection techniques have been proposed
fies the advanced detection approaches into four categories and to detect numerous faults in PV systems [23]–[41]. A brief
analyzes each approach in detail model-based difference mea- overview of these techniques along with its merits and demerits
surement (MBDM), real-time difference measurement (RDM), is presented in Table II.
output signal analysis (OSA), and machine learning techniques
(MLT). Table I specifies the fundamental differences between B. Real-Time Difference Measurement
each detection approach, technology requirement, and type of
RDM is a similar approach derived from the basic MBDM
thresholds required detection.
technique. Unlike MBDM approach, real-time PV modeling
is not essential for RDM-based detection techniques. Instead,
A. Model-Based Difference Measurement in RDM, the PV modeling or real-time experimentations are
utilized to set the threshold limits for the detection variables in
PV modeling is the virtual realization of PV systems to ac-
an offline mode. The computed real-time values for detection
curately emulate their real-time operating characteristics [22].
variables are then compared with the previously set threshold
MBDM techniques work on the principle that faulty operation
limits to detect the fault. If “Vdet th m ax ” and “Vdet th m in ” are
of PV systems brings about a change in the real-time output
the upper and lower threshold limits of the detection variable,
performance parameters, such as power delivered, operating
“Vdet ,” the operating status of the PV system is identified by
voltage, and operating current. The detection rule in MBDM
evaluating the following condition:
techniques is to compare the expected value of these perfor-  
mance parameters with the sensed real-time values. The sensed Normal; Vdet th m ax ≤ Vdet ≤ Vdet th m in
real-time instantaneous data, “Vdet real ” is compared with the Status = .
Faulty; otherwise
model predicted ideal parameter value, “Vdet sim ” to identify
(2)
the fault condition; and can be mathematically represented as
For clarity, fault detection using the RDM approach is ex-
  plained with the help of flowchart depicted in Fig. 3. RDM
Faulty; Vdet real = ±k Vdet sim techniques have been widely used for real-time fault detection
Status = . (1)
Normal; Vdet real = Vdet sim in PV systems [42]–[54]; and different methods that employ
516 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 9, NO. 2, MARCH 2019

TABLE II
OVERVIEW ON MODEL-BASED DIFFERENCE MEASUREMENT TECHNIQUES FOR FAULT DETECTION

RDM approach for fault detection are reviewed and presented in the output signals; while, arc faults cause continuous oscilla-
in Table III. tions and distortions in the output current and voltage waveforms
of the inverter. OSA techniques rely on suitable frequency/time
domain analysis to detect the anomalies present in the sampled
C. Output Signal Analysis
signals in order to identify a particular fault. Once, the threshold
Usually, irrespective of the environmental conditions, the limits for decision making are obtained using theoretical cal-
terminal characteristic of any PV system is affected due to the culations/analysis, the violation of these limits can be checked
occurrence of faults. For instance, shade faults, ground faults, to detect the presence of a fault. For a better understanding,
and short-circuit faults are associated with catastrophic changes the flowchart for the OSA approach is presented in Fig. 4.
PILLAI et al.: COMPARATIVE EVALUATION OF ADVANCED FAULT DETECTION APPROACHES FOR PV SYSTEMS 517

Fig. 3. Flowchart for real-time difference measurement approach.


Fig. 4. Flowchart for output signal analysis approach.

An outline of researches based on the OSA approach available in


detection technique can be integrated with PV installations. On
the literature [55]–[68] are compiled and presented in Table IV.
the other hand, the detection capability test critically reviews the
reliability of each fault detection technique at different unique
D. Machine Learning Techniques
operating conditions of PV systems.
To improve the potential ways of fault detection in PV sys-
tems, machine learning techniques, such as artificial neural net- A. Practical Fitness Analysis
works (ANNs), fuzzy logic systems, etc., are also deployed by
In order to evaluate the practical fitness of each fault detection
the researchers. In MLT-based fault detection, learning algo-
technique, three important performance parameters are assessed
rithms are trained using experimental or model predicted PV
number of faults detected, number of steps involved in the detec-
data to learn the relationship between input and output parame-
tion process, and additional sensor and component requirement.
ters of a PV system. However, in addition to the normal PV data,
The results obtained for various MBDM, RDM, OSA, and MLT
anomalous data due to fault occurrences are also required for
fault detection techniques are illustrated in Fig. 6(a)–(d), re-
accurate training. An accurately trained artificial model can pre-
spectively. Note that the number of additional sensors/devices
cisely predict the real-time performance parameters (both ideal
indicated in the results shows the minimum number required
and faulty) once; the instantaneous irradiation levels and tem-
for a single PV string. From the data presented, the following
perature are sensed and fed as inputs to the model. Hence, any
inferences are arrived.
faulty state of the PV array can easily be identified using simple
1) Use of the real-time PV modeling in MBDM approaches
error evaluations. The flowchart for MLT-based fault detection
necessitates numerous additional components. For in-
is presented in Fig. 5 and a brief survey and an analysis on the
stance, as illustrated in Fig. 6(a), all MBDM techniques
various research works carried out in this domain [69]–[76] is
presented in [23]–[41] require at least one additional com-
provided in Table V.
ponent for accurate real-time PV modeling, i.e., irradiance
sensor for illumination measurement. In addition, except
IV. COMPATIBILITY STUDY
[33], [36], and [39], all MBDM techniques require tem-
After meaningful understanding on various advanced fault de- perature measurement as well. Moreover, utilization of
tection approaches, it is intended to evaluate the compatibility of weather stations and data acquisition systems [23]–[25],
these approaches for practical implementation. For which, two [29], [38], [40], and [41] makes MBDM schemes very
different analysis have been carried out practical fitness analy- complex and difficult to integrate with PV installations.
sis and detection capability test. The review based on practical 2) Unlike MBDM, RDM techniques do not require addi-
fitness helps to analyze the procedural complexity involved in tional components/sensors for online PV modeling. In-
various detection techniques and the easiness at which each stead, sensor requirement for RDM techniques depend
518 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 9, NO. 2, MARCH 2019

TABLE III
OVERVIEW ON REAL TIME DIFFERENCE MEASUREMENT TECHNIQUES FOR FAULT DETECTION

on the type of detection variables. For example, tech- analysis that just requires an additional software platform
niques proposed in [46]–[48], and [52] require no addi- for feature extraction of output signals. Therefore, most
tional sensors since, fault detection is based on the MPP OSA techniques need only a single feature extraction de-
voltage and current data alone; that are available at the vice for practical implementation [55], [56], [58]–[60],
converter terminals of MPPT-based PV systems. How- [62], [63], [65], [66], [68]. Thus, OSA techniques are
ever, to further improve the accuracy, some specific RDM mostly compact, commercially effective, and practically
techniques use additional irradiance and temperature mea- feasible.
surements [42]–[45] too. Further, techniques explained 4) On the other hand, machine learning techniques proposed
in [49], [51], [53], and [54] need both data acquisition for fault detection needs the instantaneous values of ir-
system and irradiance measurement. Altogether, in con- radiance and temperature to classify the fault [69]–[76].
trast to MBDM, owing to the less component require- Apart from the sensor requirement, the prime disadvan-
ment, RDM techniques are more feasible for practical tage of MLT-based fault detection is its site specific design
implementation. procedure; that makes the detection algorithm incompat-
3) Compared to both MBDM and RDM approach, the addi- ible for large-scale PV installations. Thus, lack universal
tional sensor/component requirement for OSA techniques fault detection rules restrict the applicability of the MLT
is very less. This is because OSA techniques rely on signal approach for fault detection in PV systems.
PILLAI et al.: COMPARATIVE EVALUATION OF ADVANCED FAULT DETECTION APPROACHES FOR PV SYSTEMS 519

TABLE IV
OVERVIEW ON OUTPUT SIGNAL ANALYSIS TECHNIQUES FOR FAULT DETECTION

To summarize, the analysis conducted convey that, compared following OSA and RDM approaches are realistic and seem to
to MBDM and MLT-based fault detection techniques, OSA and be efficient for practical implementation.
RDM techniques are easy to integrate with already existing
as well as new PV installations with minimal modifications to B. Detection Capability Test
the system architecture. In this paper, techniques proposed in For detection capability test, LL, LG, and arc faults that are
[46]–[48], [52], [55], [56], [59], [60], [63], [65], [66], and [68] most common in PV arrays are considered and the performance
520 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 9, NO. 2, MARCH 2019

TABLE V
OVERVIEW ON MACHINE LEARNING TECHNIQUES FOR FAULT DETECTION

of each fault detection technique available for LL, LG, and arc
fault detection is evaluated in the context of existing protection
challenges.
1) LL and LG Fault Analysis: Any advanced method pro-
posed for LL and LG fault detection must be reliable to counter
act the conventional protection challenges involved with the
standard protection devices. Hence, the capability of advanced
LL/LG fault detection techniques to detect faults is evaluated
in the context of fault impedance, irradiation levels, percentage
mismatch, failure probability, applicability, and level of protec-
tion achieved. Here, fault impedance refers to the impedance
offered to the fault current flow and percentage mismatch indi-
cates the percentage of healthy modules in a faulty PV string. It
is important to note that, as per [1], [10], and [11], an ideal
LL/LG fault detection technique must be capable to detect
LL/LG faults with high fault impedance, in low irradiation lev-
els, and at low mismatch levels. Further, differentiating LL/LG
faults from PS conditions (PSCs) is equally important; since,
both possess the same signature. Therefore, the detection capa-
bility of each LL/LG fault detection technique is evaluated with
respect to the aforementioned constraints and the investigation
outcomes are elucidated in Table VI. It is evident from the re-
sults that RDM and OSA techniques are more realistic in case
of LL fault detection; since, most approaches are capable for
Fig. 5. Flowchart for machine learning techniques for fault detection. fault detection in low illumination levels, low mismatch levels
PILLAI et al.: COMPARATIVE EVALUATION OF ADVANCED FAULT DETECTION APPROACHES FOR PV SYSTEMS 521

Fig. 6. Practical fitness analysis on advanced fault detection techniques.


522 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 9, NO. 2, MARCH 2019

TABLE VI
DETECTION CAPABILITY TEST FOR LL/LG FAULT DETECTION

and PSCs. Similarly, OSA techniques perform well in LG fault impedance. It is also worth to mention that recently conceptual-
detection too; whereas, RDM and MLT approaches are yet to ized MPPT-based fault detection approach propounded in [77]
be adopted for LG fault detection. From the analysis given in is highly reliable in all aforementioned aspects and opens up
Table VI, the following conclusions can be arrived: new possibilities in the LL/LG fault detection.
1) for LL fault detection, RDM techniques [51], [56], and 2) Arc Fault Analysis: Protection of PV arrays from arc
[57], and OSA techniques [61], [67], [68], and [73] are faults using conventional AFCIs are not reliable since its op-
efficient at mismatch levels below 25%, low illumination eration is primarily affected by the interference caused due to
levels even below 500 W/m2 and in PSCs; inverter switching; that leads to frequent nuisance tripping [10].
2) for LG fault detection, OSA technique in [74] seems to Further, series arcs can be easily extinguished by opening the in-
provide the most reliable solution; verter terminals; while doing so in case of parallel arcs increases
3) all MBDM and MLT-based detection techniques except the current flow through the arc path; as the PV array enters into
[76] fail to detect LL/LG faults in PSCs and open-circuit conditions and operate at a higher voltage [10],
4) detecting LL/LG faults with high fault impedance still [11]. Therefore, any arc fault detection technique recommended
remains as a challenging task and only OSA methods [61], for PV systems must be proficient to discriminate series and par-
[63], [67], [68], and [74] are capable of high impedance allel arc faults to ensure safe operation. Since dc arc faults are
fault detection. characterized by signatures in both frequency and time domain,
Overall, RDM and OSA approaches are found to be pro- analytical methods are usually adopted for PV arc fault detec-
ficient in LL/LG fault detection. However, irrespective of the tion. In the literature, compared to other faults, very few arc fault
approach, a notable drawback of advanced LL/LG fault detec- detection techniques are available. To date, only OSA-based
tion techniques is the inability to detect faults with high fault fault detection techniques are applied for arc fault detection in
PILLAI et al.: COMPARATIVE EVALUATION OF ADVANCED FAULT DETECTION APPROACHES FOR PV SYSTEMS 523

TABLE VII
DETECTION CAPABILITY TEST FOR ARC FAULT DETECTION

Fig. 7. Detection capability of advanced detection approaches. (a) LG faults. (b) LL faults. (c) Arc faults.

PV systems [59], [63], [64], and [66]. In order to evaluate the comparative study based on the merits and demerits of various
compatibility of advanced arc fault detection techniques, each approaches is carried out in this section. Further, the implica-
technique is qualitatively gauged with respect to capability to tions from the detailed survey conducted are utilized to arrive at
differentiate between series and parallel arc faults, domain used a generalized performance evaluation of various advanced fault
for fault detection, dependency with inverter operation, and re- detection approaches in a common platform.
quirement of manually set threshold limits. The review results
obtained after detailed evaluations are illustrated in Table VII.
Further, the evaluations exemplify that, none of the arc fault A. Merits and Demerits of Advanced Detection Approaches
detection techniques possess the capability to detect parallel
As previously indicated, MBDM, RDM, and MLT-based fault
arcs. Further, most methods cannot detect arcs universally as
detection techniques rely on the PV modeling (either offline or
they require system specific and manually set threshold limits
online) for fault detection. Hence, selection of PV model plays
for fault detection [63], [64], and [66]. Although the detection
a crucial role in imparting accuracy to these approaches. The
technique suggested in [66] based on modified Tsallis entropy
PV model applied must be accurate to predict precise detection
is not suitable for parallel arc detection, it can automatically set
variables in all operating conditions. Therefore, accuracy of
the threshold limits and is not influenced by inverter switching.
MBDM, RDM, and MLT approaches is reliant on the reliability
Therefore, compared to other detection techniques, [66] is ver-
of the parameter extraction technique used to define the PV
satile in various aspects and is expected to be the most viable
model; which in fact imparts accuracy to the model [78]. Further,
method available for arc fault detection. In summary, PV arc
RDM, OSA, and MLT approaches can aid rapid fault detection
fault detection demands more research attention and potential
compared to MBDM as real-time PV modeling is not required
techniques that can detect both series and parallel arc faults are
for detection; that significantly reduces the number of detection
yet be evolved. For clarity, the findings on detection capability
steps. It is important to note that, although noise in sampling
are consolidated and are pictorially represented in Fig. 7(a)–(c),
signals affects the accuracy, OSA techniques are commercially
respectively, for LG, LL, and arc fault detection.
effective, product oriented, and accurate. On the other hand,
ML-based fault detection techniques do not require threshold
V. OVERALL REVIEW ON ADVANCED FAULT DETECTION
limits for fault detection and if properly trained classification
APPROACHES and localization of faults becomes very easy. However, noise in
The investigations carried out in the previous section evalu- the training data limits the accuracy of the trained model. To
ated the compatibility of each fault detection technique in the conclude, a brief insight and concluding remarks on the merits
literature. In order to enhance the analysis, a comprehensive and demerits of various approaches are given in Table VIII.
524 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 9, NO. 2, MARCH 2019

TABLE VIII
ADVANTAGES AND DISADVANTAGES OF ADVANCED FAULT DETECTION APPROACHES

Fig. 8. Comparative evaluation of advanced fault detection approaches with conventional protection schemes.
PILLAI et al.: COMPARATIVE EVALUATION OF ADVANCED FAULT DETECTION APPROACHES FOR PV SYSTEMS 525

B. Generalized Performance Evaluation with respect to six different performance parameters. Based on
the critical implications presented in this paper, the following
The implications drawn from the extensive analysis per-
formed in this paper are used to compare the performance of conclusive remarks and suggestions that are likely to enhance
fault detection possibilities in PV systems are presented as
advanced fault detection approaches in a general platform. Here,
follows.
each advanced detection approach is compared with the con-
ventional protection schemes with respect to six performance 1) RDM and OSA-based fault detection techniques are ef-
ficient and can be integrated to any PV systems with
parameters that are selected based on the implications drawn
minimal modifications to the system architecture. How-
from the investigations performed in the previous sections envi-
ronmental impact, procedural complexity, components required, ever, component requirement and computational burden
are major disadvantages of these approaches. Therefore,
integration complexity, detection time, and effect of noise. The
performance evaluation of each approach based on the above- researches confined to improve these assets are expected
mentioned performance parameters is prepared in the form of to give impetus for advanced detection techniques in near
future.
spider diagrams; where every parameter is qualitatively assessed
to provide a fair taxation on the approach. The assessment is 2) Among advanced detection techniques, for protection
presented with the help of four spider diagrams illustrated in against LL and LG faults, RDM and OSA techniques
are feasible to implement irrespective of the system-type
Fig. 8(a)–(d), respectively; with each axis in the spider chart rep-
resents a performance parameter that is qualitatively assessed and rating. For arc fault detection, OSA techniques are
in a range from low to high starting from the center point. Note realistic; but necessitate future research advancements to
reduce the feature extraction time.
that the chart can be understood in the following way; lower the
area covered, better the method is. 3) Arc fault detection is less studied in the literature com-
pared to other PV faults. Hence, a reliable arc fault detec-
As previously studied, the primary challenge in advanced
tion technique that can detect both series and parallel arc
fault detection is to cope up with the unique operating char-
acteristics of PV systems in different environmental condi- faults is essential for PV systems.
4) Differentiating parallel and series arcs is a major concern
tions. From Fig. 8, it is evident that all advanced fault detec-
in arc fault detection and mitigation. For which, the drop
tion approaches perform better than the conventional protection
schemes in terms of environmental impact. To be more precise, in array current and voltage due to parallel arcs can be
utilized; which does not happen in the case of series arc
except for the techniques presented in [34], [37], [39], [44],
faults.
[48], [54], [59], and [61], this observation holds true. Similar
analysis has also been conducted for all other performance pa- 5) Setting thresholds for fault detection using PV model is
inevitable; hence, extended investigations to enrich PV
rameters to realize the remaining spider diagram and it can be
visualized that all advanced approaches are practically feasible modeling and parameter estimation of different commer-
with respect to integration complexity and components require- cially available PV modules perceive utmost importance.
Further, a PV model that can predict precise detection
ment. But, compared to conventional protection schemes, most
advanced detection approaches have elevated component re- variables at all operating conditions must be selected for
quirement and high procedural complexity. However, even with setting threshold limits.
6) Conventional protection schemes that rely on fault cur-
its elevated requirements, advanced approaches are immensely
capable to counteract the conventional protection challenges. rent magnitudes can no longer protect PV systems in all
For example, RDM techniques [49]–[51] and [53] and OSA operating conditions. Therefore, to enhance the reliabil-
ity and safety of PV systems, advanced fault detection
techniques [65], [67], and [68] need additional components
for implementation; but, can guarantee reliable fault detection. techniques can be installed as a supplementary protection
scheme along with the conventional protection devices.
Overall, among the various advanced fault detection approaches,
RDM and OSA approach possess uniform characteristics (see
Fig. 8) except for its sensitiveness to noise levels. Therefore, ACKNOWLEDGMENT
these approaches can be presumed to be the most efficient and
The authors would like thank the reviewers and the Ed-
feasible techniques available for real-time fault detection in PV
itor, Prof. G. Spagnuolo, for their valuable comments and
systems.
suggestions.
VI. CONCLUSIVE REMARKS AND SUGGESTIONS
REFERENCES
This paper briefly discussed the necessity of advanced fault
detection approaches in PV systems. The advanced fault detec- [1] Y. Zhao, J. de Palma, J. Mosesian, R. Lyon, and B. Lehman, “Line-line
fault analysis and protection challenges in solar photovoltaic arrays,” IEEE
tion approaches were classified into four categories and were Trans. Ind. Electron., vol. 60, no. 9, pp. 3784–3795, Sep. 2013.
comprehensively studied. In addition, two assessment tests [2] C. Huang et al., “A prediction model-guided Jaya algorithm for the PV
namely compatibility analysis and detection capability test have system maximum power point tracking,” IEEE Trans. Sustain. Energy,
vol. 9, no. 1, pp. 45–55, Jan. 2018.
also been carried out to evaluate the prominence of advanced [3] H. S. Sahu and S. K. Nayak, “Extraction of maximum power from a
detection techniques. Furthermore, the performance of these ap- PV array under nonuniform irradiation conditions,” IEEE Trans. Electron
proaches is compared with the conventional protection schemes Devices, vol. 63, no. 12, pp. 4825–4831, Dec. 2016.
526 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 9, NO. 2, MARCH 2019

[4] D. S. Pillai, J. P. Ram, M. S. S. Nihanth, and N. Rajasekar, “A sim- [30] M. Hosseinzadeh and F. R. Salmasi, “Determination of maximum so-
ple, sensorless and fixed reconfiguration scheme for maximum power en- lar power under shading and converter faults—a prerequisite for failure-
hancement in PV systems.” Energy Conv. Manag., vol. 172, pp. 402–417, tolerant power management systems,” Simul. Model. Pract. Theory,
Sep 2018. vol. 62, pp. 14–30, Mar. 2016.
[5] E. Scolari, L. Reyes, F. Sossan, and M. Paolone, “A comprehensive as- [31] C. Ventura and G. M. Tina, “Utility scale photovoltaic plant indices and
sessment of the short-term uncertainty of grid-connected PV systems,” models for on-line monitoring and fault detection purposes,” Elect. Power
IEEE Trans. Sustain. Energy, vol. 9, no. 3, pp. 1458–1467, Jul. 2018. Syst. Res., vol. 136, pp. 43–56, Jul. 2016.
[6] Y. Wang and B. Ren, “Fault ride-through enhancement for grid-tied PV [32] M. Dhimish and V. Holmes, “Fault detection algorithm for grid-connected
systems with robust control,” IEEE Trans. Ind. Electron., vol. 65, no. 3, photovoltaic plants,” Solar Energy, vol. 137, pp. 236–245, Nov. 2016.
pp. 2302–2312, Aug. 2017. [33] M. Bressan, Y. El Basri, A. G. Galeano, and C. Alonso, “A shadow fault
[7] Article 690: Solar PV Systems, National Electrical Code, ANSI/NFPA-70, detection method based on the standard error analysis of IV curves,”
Nat. Fire Protection Assoc., Inc., Quincy, MA, USA, 2011. Renew. Energy, vol. 99, pp. 1181–1190, Dec. 2016.
[8] IEEE Guide for Terrestrial PV Power System Safety, IEEE Std. 1374. [34] A. Umana and A. P. S. Meliopoulos, “Detection of cell-level fault con-
1998. ditions within a photovoltaic array system,” in Proc. IEEE Power Energy
[9] Installation and Safety Requirements for PV Generators, IEC Std. 62548, Soc., 2016, pp. 1–5.
2013. [35] W. Wang et al., “Fault diagnosis of photovoltaic panels using dynamic
[10] M. K. Alam, F. Khan, J. Johnson, and J. Flicker, “A comprehensive re- current–voltage characteristics.” IEEE Trans. Power Electron., vol. 31,
view of catastrophic faults in PV arrays,” Types detection and mitigation no. 2, pp. 1588–1599, Feb. 2016.
techniques,” IEEE J. Photovolt., vol. 5, no. 3, pp. 982–997, May 2015. [36] T. Andrianajaina, E. J. R. Sambatra, C. B. Andrianirina, T. D. Razafima-
[11] D. S. Pillai and N. Rajasekar, “A comprehensive review on protection hefa, and N. Heraud, “PV fault detection using the least squares method,”
challenges and fault diagnosis in PV systems,” Renew. Sustain. Energy in Proc. IEEE Eur. Conf. Power Electron. Appl., 2016, pp. 846–851.
Rev., vol. 91, pp. 18–40, Aug. 2018. [37] A. Belaout, F. Krim, and A. Mellit, “Neuro-fuzzy classifier for fault detec-
[12] Y. Zhao, “Fault analysis in solar photovoltaic arrays,” Ph.D. diss., North- tion and classification in photovoltaic module,” in Proc. IEEE Int. Conf.
eastern Univ., Boston, MA, USA, 2011. Model. Identification Control, 2016, pp. 144–149.
[13] W. I. Bower and J. C. Wiles, “Analysis of grounded and ungrounded [38] E. Garoudja et al., “Statistical fault detection in photovoltaic systems,”
photovoltaic systems.” in Proc. IEEE Photovolt. Spec. Conf., 1994, vol. 1, Solar Energy, vol. 150, pp. 485–499, Jul. 2017.
pp. 809–812. [39] F. Harrou, Y. Sun, B. Taghezouit, A. Saidi, and M. E. Hamlati, “Re-
[14] S. E. Forman, “Performance of experimental terrestrial photovoltaic mod- liable fault detection and diagnosis of photovoltaic systems based on
ules,” IEEE Trans. Rel. vol. 31, no. 3, pp. 235–245, Aug. 1982. statistical monitoring approaches.” Renew. Energy, vol. 116, pp. 22–37,
[15] R. Kase and S. Nishikawa, “Fault detection of bypass circuit of PV Feb. 2018.
module—Detection technology of open circuit fault location,” in Proc. [40] M. Dhimish, V. Holmes, B. Mehrdadi, and M. Dales, “Simultaneous fault
IEEE Int. Conf. Elect. Mach. Syst., 2016, pp. 1–4. detection algorithm for grid-connected photovoltaic plants,” IET Renew.
[16] E. D. Spooner and N. Wilmot, “Safety issues, arcing and fusing in PV Power Gener., vol. 12, pp. 1565–1575, Jul 2017.
arrays,” in Proc. ISESC, 2008. [41] M. Dhimish, V. Holmes, B. Mehrdadi, and M. Dales, “Multi-layer pho-
[17] C. Strobl and P. Meckler, “Arc faults in photovoltaic systems.” in Proc. tovoltaic fault detection algorithm.” High Voltage, vol. 4, pp. 244–252,
IEEE Human Corneal Epithelial Cells, 2010, pp. 1–7. May 2017.
[18] G. Dhere and N. S. Shiradkar, “Fire hazard and other safety concerns of [42] T. Shimakage, K. Nishioka, H. Yamane, M. Nagura, and M. Kudo, “De-
photovoltaic systems,” J. Photon. Energy, vol. 2, no. 1, pp. 22006–22006, velopment of fault detection system in PV system,” in Proc. IEEE 33rd
Dec. 2012. Int. Telecommun. Energy Conf., 2011, pp. 1–5.
[19] E. Koutroulis and F. Blaabjerg, “A new technique for tracking the global [43] X. Xu, H. Wang, and Y. Zuo, “Method for diagnosing photovoltaic array
maximum power point of PV arrays operating under partial-shading con- fault in solar photovoltaic system,” in Proc. IEEE Asia-Pac. Power Energy
ditions.” IEEE J. Photovolt., vol. 2, no. 2, pp. 184–190, Apr. 2012. Eng. Conf., 2011, pp. 1–5.
[20] S. V. Dhople, A. Davoudi, A. D. Domı́nguez-Garcı́a, and P. L. Chapman, [44] M. Omana, D. Rossi, G. Collepalumbo, C. Metra, and F. Lombardi, “Faults
“A unified approach to reliability assessment of multiphase DC–DC con- affecting the control blocks of PV arrays and techniques for their concur-
verters in photovoltaic energy conversion systems,” IEEE Trans. Power rent detection,” in Proc. IEEE Defect Fault Tolerance VLSI Nanotechnol.
Electron., vol. 27, no. 2, pp. 739–751, Feb. 2012. Syst., 2012, pp. 199–204.
[21] J. Johnson et al., “Photovoltaic DC arc fault detector testing at San- [45] N. Gokmen, E. Karatepe, B. Celik, and S. Silvestre, “Simple diagnos-
dia National Laboratories,” in Proc. IEEE Photovolt. Spec. Conf., 2011, tic approach for determining of faulted PV modules in string based PV
pp. 3614–3619. arrays,” Solar Energy, vol. 86, no. 11, pp. 3364–3377, Nov. 2012.
[22] D. S. Pillai, N. Rajasekar, J. P. Ram, and V. K. Chinnaiyan, “Design [46] J. Solórzano and M. A. Egido, “Automatic fault diagnosis in PV systems
and testing of two phase array reconfiguration procedure for maximizing with distributed MPPT,” Energy Convers. Manage., vol. 76, pp. 925–934,
power in solar PV systems under partial shade conditions (PSC),” Energy Dec. 2013.
Conv. Manag., vol. 178, pp. 92–110, Dec. 2018. [47] S. Silvestre, M. A. da Silva, A. Chouder, D. Guasch, and E. Karatepe,
[23] W. Chine, A. Mellit, A. M. Pavan, and S. A. Kalogirou, “Fault detection “New procedure for fault detection in grid connected PV systems based
method for grid-connected photovoltaic plants,” Renew. Energy, vol. 66, on the evaluation of current and voltage indicators,” Energy Convers.
pp. 99–110, Jun. 2014. Manage., vol. 86, pp. 241–249, Oct. 2014.
[24] S. Silvestre, A. Chouder, and E. Karatepe, “Automatic fault detection [48] X. Lin, Y. Wang, M. Pedram, J. Kim, and N. Chang, “Designing fault-
in grid connected PV systems,” Solar Energy, vol. 94, pp. 119–127, tolerant photovoltaic systems,” IEEE Des. Test, vol. 31, no. 3, pp. 76–84,
Aug. 2013. Jun. 2014.
[25] M. Davarifar, A. Rabhi, A. Elhajjaji, and M. Dahmane, “New method for [49] Y. Hu et al., “Online two-section PV array fault diagnosis with optimized
fault detection of PV panels in domestic applications,” in Proc. IEEE Int. voltage sensor locations,” IEEE Trans. Ind. Electron., vol. 62, no. 11,
Conf. Semantic Comput., 2013, pp. 727–732. pp. 7237–7246, Nov. 2015.
[26] L. Bonsignore, M. Davarifar, A. Rabhi, G. M. Tina, and A. Elhajjaji, [50] M. Alajmi and I. Abdel-Qader, “Fault detection and localization in solar
“Neuro-fuzzy fault detection method for photovoltaic systems,” Energy photovoltaic arrays using the current-voltage sensing framework,” in Proc.
Procedia, vol. 62, pp. 431–441, Jan. 2014. IEEE Elect. Inf. Technol., pp. 307–312, 2016.
[27] M. Tadj, K. Benmouiza, A. Cheknane, and S. Silvestre, “Improving the [51] R. Hariharan, M. Chakkarapani, G. S. Ilango, and C. Nagamani, “A
performance of PV systems by faults detection using GISTEL approach,” method to detect photovoltaic array faults and partial shading in PV sys-
Energy Convers. Manage., vol. 80, pp. 298–304, Apr. 2014. tems,” IEEE J. Photovolt., vol. 6, no. 5, pp. 1278–1285, Sep 2016.
[28] W. Chine, A. Mellit, A. M. Pavan, and V. Lughi, “Fault diagnosis in [52] M. Bressan, Y. El-Basri, and C. Alonso, “A new method for fault detection
photovoltaic arrays,” in Proc. IEEE Int. Conf. Clean Elect. Power, 2015, and identification of shadows based on electrical signature of defects,” in
pp. 67–72. Proc. IEEE 17th Eur. Conf. Power Electron. Appl., 2015, pp. 1–8.
[29] R. Platon, J. Martel, N. Woodruff, and T. Y. Chau, “Online fault detection [53] E. Garoudja, K. Kara, A. Chouder, S. Silvestre, and S. Kichou, “Efficient
in PV systems,” IEEE Trans. Sustain. Energy, vol. 6, no. 4, pp. 1200–1207, fault detection and diagnosis procedure for photovoltaic systems,” in Proc.
Oct. 2015. IEEE 8th Int. Conf. Model. Identification Control, 2016, pp. 851–856.
PILLAI et al.: COMPARATIVE EVALUATION OF ADVANCED FAULT DETECTION APPROACHES FOR PV SYSTEMS 527

[54] A. Dhoke, R. Sharma, and T. K. Saha, “PV module degradation analysis [77] D. S. Pillai and N. Rajasekar, “An MPPT based sensorless line-line and
and impact on settings of overcurrent protection devices,” Solar Energy, line-ground fault detection technique for PV systems,” IEEE Trans. Power
vol. 160, pp. 360–367, Jan 2018. Electron., to be published, doi: 10.1109/TPEL.2018.2884292.
[55] Y. Zhao, B. Lehman, R. Ball, J. Mosesian, and J. de Palma, “Outlier [78] D. S. Pillai and N. Rajasekar, “Metaheuristic algorithms for PV parameter
detection rules for fault detection in solar photovoltaic arrays.” in Proc. identification: A comprehensive review with an application to threshold
IEEE 28th Ann. IEEE Appl. Power Electron. Conf. Expo., 2013, pp. 2913– setting for fault detection in PV systems,” Renew. Sustain. Energy Rev.,
2920. vol. 82, no. 3, pp. 3503–3525, Feb. 2018.
[56] Y. Zhao et al., “Fault experiments in a commercial-scale PV laboratory and
fault detection using local outlier factor,” in Proc. IEEE 40th Photovolt.
Spec. Conf., 2014, pp. 3398–3403.
[57] G. Wang, C. C. Youn, and A. M. Stankovic, “DC-side high impedance
ground fault detection for transformer less single-phase PV systems,” in Dhanup S. Pillai received the B.Tech. degree in elec-
Proc. IEEE North Amer. Power Symp., 2015, pp. 1–6. trical and electronics engineering and the M.E. de-
[58] I. S. Kim, “On-line fault detection algorithm of a photovoltaic system using gree in power electronics and drives from Mahatma
wavelet transform,” Solar Energy, vol. 126, pp. 137–145, Mar. 2016. Gandhi University, Kottayam, India, and Anna Uni-
[59] N. Georgijevic, M. V. Jankovic, S. Srdic, and Z. Radakovic, “The detection versity, Chennai, India, respectively. He is working
of series arc fault in photovoltaic systems based on the arc current entropy,” toward the Ph.D. degree in electrical engineering at
IEEE Trans. Power Electron., vol. 31, no. 8, pp. 5917–5930, Aug. 2016. the Vellore Institute of Technology, Vellore, India.
[60] K. A. Kim, G. Seo, B. Cho, and P. T. Krein, “Photovoltaic hot-spot de- From 2011 to 2016, he was an Assistant Professor
tection for solar panel substrings using AC parameter characterization,” with the Musaliar College of Engineering and Tech-
IEEE Trans. Power Electron., vol. 31, no. 2, pp. 1121–1130, Feb. 2016. nology, Pathanamthitta, India. His interests include
[61] L. Chen, S. Li, and X. Wang, “Quickest fault detection in photovoltaic protection and fault detection in PV systems, Line-
systems,” IEEE Trans. Smart Grid, vol. 9, no. 3, pp. 1835–1847, May 2018. Line and Line-Ground fault detection in PV systems, optimization techniques,
[62] Z. Yi and A. H. Etemadi, “Fault detection for photovoltaic systems based power electronics and applications of power electronics to renewable energy
on multi-resolution signal decomposition and fuzzy inference systems,” systems.
IEEE Trans. Smart Grid, vol. 8, no. 3, pp. 1274–1283, May 2017.
[63] S. Chen and X. Li, “PV series arc fault recognition under different working
conditions with joint detection method,” in Proc. IEEE Elect. Contacts,
2016, pp. 25–32.
[64] H. Zhu, Z. Wang, and R. S. Balog, “Real time arc fault detection in PV Frede Blaabjerg (S’86–M’88–SM’97–F’03) re-
systems using wavelet decomposition,” in Proc. IEEE 43rd Photovolt. ceived the Ph.D. degree in electrical engineering from
Spec. Conf., 2016, pp. 1761–1766. Aalborg University, Aalborg, Denmark, in 1995.
[65] S. Dhar, R. K. Patnaik, and P. K. Dash, “Fault detection and location of From 1987 to 1988, he was with ABB-Scandia,
photovoltaic based DC microgrid using differential protection strategy,” Randers, Denmark. In 1992, he was an Assistant Pro-
IEEE Trans. Smart Grid, vol. 9, no. 5, pp. 4303–4312, Sep. 2018. fessor, an Associate Professor in 1996, and a Full
[66] Q. Xiong, S. Ji, L. Zhu, L. Zhong, and Y. Liu, “A novel DC arc fault Professor of power electronics and drives in 1998.
detection method based on electromagnetic radiation signal,” IEEE Trans. In 2017, he was a Villum Investigator. From 1987 to
Plasma Sci., vol. 45, no. 3, pp. 472–478, Mar. 2017. 1988, he was with ABB-Scandia, Randers. He has
[67] B. P. Kumar, G. S. Ilango, M. J. B. Reddy, and N. Chilakapati, “Online fault authored or coauthored more than 450 journal papers
detection and diagnosis in photovoltaic systems using wavelet packets,” in the fields of power electronics and its applications.
IEEE J. Photovolt., vol. 8, no. 1, pp. 257–265, Jan 2018. He is the Co-Author of two monographs and the Editor of six books in power
[68] S. Roy, M. K. Alam, F. Khan, J. Johnson, and J. Flicker, “An irradiance- electronics and its applications. His research interests include power electronics
independent, robust ground-fault detection scheme for PV Arrays based and its applications such as in wind turbines, PV systems, reliability, harmonics,
on spread spectrum time-domain reflectometry (SSTDR),” IEEE Trans. and adjustable speed drives.
Power Electron., vol. 33, no. 8, pp. 7046–7057, Aug. 2018. Dr. Blaabjerg was the recipient of 18 IEEE Prize Paper Awards, the IEEE
[69] Y. Zhao et al., “Decision tree-based fault detection and classification in PELS Distinguished Service Award in 2009, the EPE-PEMC Council Award
solar photovoltaic arrays,” in Proc. IEEE 27th Annu. IEEE Appl. Power in 2010, the IEEE William E. Newell Power Electronics Award 2014, and the
Electron. Conf. Expo., 2012, pp. 93–99. Villum Kann Rasmussen Research Award in 2014. He was the Editor-in-Chief
[70] M. N. Akram and S. Lotfifard, “Modeling and health monitoring of DC of the IEEE TRANSACTIONS ON POWER ELECTRONICS from 2006 to 2012. He
side of photovoltaic array,” IEEE Trans. Sustain. Energy, vol. 6, no. 4, was a Distinguished Lecturer for the IEEE Power Electronics Society from 2005
pp. 1245–1253, Oct. 2015. to 2007 and the IEEE Industry Applications Society from 2010 to 2011 as well
[71] E. Karatepe and T. Hiyama, “Controlling of artificial neural network for as 2017 to present. In 2017, he became Honoris Causa at University Politehnica
fault diagnosis of photovoltaic array,” in Proc. IEEE 16th Int. Conf. Intell. Timisoara, Timisoara, Romania.
Syst. Appl. Power Syst., Sep. 2011, pp. 1–6.
[72] H. Mekki, A. Mellit, and H. Salhi, “Artificial neural network-based mod-
eling and fault detection of partial shaded photovoltaic modules,” Simul.
Model. Pract. Theory, vol. 67, pp. 1–13, Sep. 2016.
[73] L. L. Jiang and D. L. Maskell, “Automatic fault detection and diagnosis Natarajan Rajasekar (M’12) received the B.E De-
for photovoltaic systems using combined artificial neural network and gree in electrical and electronics engineering from
analytical based methods,” in Proc. IEEE Int. Joint Conf. Neural Netw., the University of Madras, Chennai, India, and the
2015, pp. 1–8. M.Tech. degree in power electronics and Drives from
[74] C. B. Jones, J. S. Stein, S. Gonzalez, and B. H. King, “Photovoltaic system SASTRA University, Thanjavur, respectively, and the
fault detection and diagnostics using laterally primed adaptive resonance Ph.D. degree in electrical engineering from the Na-
theory neural network,” in Proc. IEEE 42nd Photovolt. Spec. Conf., 2015, tional Institute of Technology, Trichy, India, in 2007.
pp. 1–6. From 2005 to 2010, he was with the Institute of
[75] K. U. Rao, A. G. Parvatikar, S. Gokul, N. Nitish, and P. Rao, “A novel Road and Transport Technology, Erode. He is cur-
fault diagnostic strategy for PV micro grid to achieve reliability centered rently a Professor with the Department of energy and
maintenance,” in Proc. IEEE 1st Int. Conf. Power Electron. Intell. Control power electronics, School of Electrical Engineering
Energy Syst., 2016, pp. 1–4. (SELECT), Vellore Institute of Technology, Vellore, India. His research interests
[76] Z. Yi and A. H. Etemadi, “Line-to-line fault detection for photovoltaic ar- include solar PV system, power electronics, application of power electronics in
rays based on multiresolution signal decomposition and two-stage support renewable energy sources, and dc–dc converters.
vector machine,” IEEE Trans. Ind. Electron., vol. 64, no. 11, pp. 8546– Dr. Rajasekar was the recipient “SOPHIA LECTURING—RESEARCH
8556, Nov. 2017. GRANT,” from Sophia University, Japan.

You might also like