Energy Conversion and Management 88 (2014) 139–152

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Review

Computational Intelligence based techniques for islanding detection

of distributed generation in distribution network: A review
J.A. Laghari a,d, H. Mokhlis a,⇑, M. Karimi a, A.H.A. Bakar b, Hasmaini Mohamad c
a
Department of Electrical Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia
b
University of Malaya Power Energy Dedicated Advanced Centre (UMPEDAC), Level 4, Wisma R&D UM, Jalan Pantai Baharu, University of Malaya, 59990 Kuala Lumpur, Malaysia
c
Faculty of Electrical Engineering, University of Technology MARA (UiTM), 40450 Shah Alam, Selangor, Malaysia
d
Department of Electrical Engineering, Quaid-e-Awam University of Engineering Science & Technology, Nawabshah 67480, Sindh, Pakistan

a r t i c l e i n f o a b s t r a c t

Article history: Accurate and fast islanding detection of distributed generation is highly important for its successful oper-
Received 28 April 2014 ation in distribution networks. Up to now, various islanding detection technique based on communica-
Accepted 9 August 2014 tion, passive, active and hybrid methods have been proposed. However, each technique suffers from
certain demerits that cause inaccuracies in islanding detection. Computational intelligence based tech-
niques, due to their robustness and flexibility in dealing with complex nonlinear systems, is an option
Keywords: that might solve this problem. This paper aims to provide a comprehensive review of computational
Islanding detection
intelligence based techniques applied for islanding detection of distributed generation. Moreover, the
Artificial neural network
Fuzzy logic control
paper compares the accuracies of computational intelligence based techniques over existing techniques
Adaptive Neuro fuzzy inference system to provide a handful of information for industries and utility researchers to determine the best method
Decision tree classifier for their respective system.
Ó 2014 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
2. Unintentional and Intentional Islanding  Causes, Effects and Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
3. Islanding detection standards and test frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
4. Islanding detection techniques and their types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
4.1. Remote islanding detection techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
4.2. Passive islanding detection techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
4.3. Active islanding detection techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
4.4. Hybrid islanding detection techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
5. Computational intelligence based techniques for islanding detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
5.1. Artificial Neural Network (ANN) based islanding detection techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
5.2. Fuzzy Logic Control (FLC) based islanding detection techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
5.3. Adaptive Neuro Fuzzy Inference System (ANFIS) Based Islanding Detection Techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
5.4. Decision tree classifier based islanding detection techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5.5. Application of other computational intelligence based techniques for islanding detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
6. Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
7. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

⇑ Corresponding author. Mobile: +60 126467936.

E-mail address: hazli@um.edu.my (H. Mokhlis).

http://dx.doi.org/10.1016/j.enconman.2014.08.024
0196-8904/Ó 2014 Elsevier Ltd. All rights reserved.
140 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152
1. Introduction
The exponential growth in electricity demand has given distrib-
uted generation (DG) technology a boost in power systems. The
International Energy Agency (IEA) has listed five major factors that
contribute to this renewed interest: developments in DG technol-
ogies, constraints on the construction of new transmission lines,
increased customer demand for higher reliable electricity, the elec-
tricity market liberalization, and concerns about climate change
[1]. All of these factors are shifting the paradigm of power systems
from centralized generation to distributed generation. However,
DG changes the nature of the distribution system from passive to
active. This has created technical problems. One of the major con-
cerns in operating DG in a distribution system is the possible
occurrence of islanding. In the islanding condition, the distribution
system connected with the DG is electrically isolated from the
main grid, yet continues to be energized by the DG [2]. Islanding
can occur due to power system imbalance caused by fault, line,
and generator outages, or any other disturbance, which can result
in the splitting of the system into some islanded networks.
When islanding occurs, the distribution network is discon-
nected from the main grid through the circuit breaker operation.
The islanded network may consist of a substation, or one or more
distribution feeders. This islanding network will be sustained if
there is sufficient generation to meet the load demand(s). Further-
more, the islanded distribution networks may continuously be
energized, which may cause potential damage to the existing
equipment, utility liability concerns, and the reduction of power
reliability and power quality. IEEE standard 1547 [3] has outlined
the condition that in the case of such situation, the DG should
detect the loss of grid connection and disconnect itself from the
distribution network within 2 s (100 cycles) of the islanding state
occurrence. Some utilities require even faster detection of less than
one second to avoid of the start of auto-recloser attempt of recon-
nection. An example of this is the Danish distribution network,
where the auto-recloser time is 500 ms (0.5 s) [4].
In order to prevent these problems, the current practice is to
disconnect the DG units when islanding occurs. However, due to
the high penetration of DG, this practice will not be an economical
option, and may result in the waste of power generation from the
DG. The islanding operation may be unintentional or intentional.
Intentional islanding of a distribution network may reduce the
congestion of transmission and distribution network, improve
the overall system performance by reducing the power losses,
and improve the voltage profiles. Hence, the islanding operation
of a distribution network may be a viable option, provided that
the various issues related to it are properly addressed. Various
researches have been conducted, and are on-going, with the pur-
pose of making islanding operation a reality [5,6]. In relation to this
development, a draft series of guides, referred to as P1547.4 draft
guide for design, the operation and integration of distributed
resource island system with electric power system has been pro-
duced [7], which will serve as a guide for practicing an intentional
islanding operation in an electrical power system.
When the distribution network is disconnected from the grid,
the first step would be to detect the islanding phenomenon. This
requires an efficient islanding detection technique to detect the
islanding event in order to operate or disconnect the DG. This is
very important, as the failure to accurately detect islanding may
result in the failure of the whole distribution network. Two factors
are important in islanding detection, which are time and accuracy
for each type of DG. Up to now, several remote, passive, active and
hybrid islanding detection techniques have been proposed. How-
ever, each technique has its merits and demerits. Hence, the
research interest shifts towards the application of computational
intelligence-based techniques for islanding detection. The compu-
tational intelligence based techniques, due to their robustness and
ability to easily deal with complex system, may be suitable for
islanding detection of DG. This paper reviews the ability of these
techniques in accurate and fast islanding detection.
The paper is organized as follows. Section 2 discusses the
causes, effects, and solution of unintentional and intentional islan-
ding. Section 3 presents the islanding detection standards and test
frame. Section 4 discusses the comparison of different existing
islanding detection techniques, along with their limitations. Sec-
tion 5 and 6 reviews computational intelligence based techniques
applications for islanding detection and discussions. Finally, the
conclusion is presented.
2. Unintentional and Intentional Islanding  Causes, Effects and
Solutions
Cascading failures and blackouts are the most significant
threats to the security of power systems. Since the previous dec-
ade, the world has encountered many power blackouts as a result
of cascading failures [8–14]. Cascading failure can be defined as a
process where one failure leads to successive failure of other ele-
ments of the grid. It may possess high risks towards the integrity
of power system network, and might causes it’s splitting into var-
ious unintentional islands. Furthermore, these unintentional
islands may encounter active or reactive power deficiency, which
may lead to frequency, angle, or voltage instability. These fre-
quency, angle, or voltage instability may further cause tripping of
other region if not controlled properly. During such condition,
the secure and reliable operation of the whole power system net-
work in an interconnected mode may be very difficult to control.
It may result in the propagation of these instabilities in other stable
parts of the network. Hence, intentional or controlled islanding has
been utilized as a preventive strategy to minimize the losses
caused by unintentional islanding [15]. Intentional islanding is
the process of intentionally splitting the grid into separate control-
lable islands [16].
Recently, the research in an intentional islanding operation of
distribution systems has increased due to economical and techni-
cal factors [17]. Intentional islanding of power systems is advanta-
geous that it separates the faulty section from the stable section of
power system. By doing this, the whole power system is split into
controlled island regions for easy handling. In such situation, each
island region should have sufficient generation to supply its loads
in order to remain operative [15]. Fig. 1 shows an example of
power system that is divided into three-island region after success-
ful splitting.
It should also be pointed out that intentional islanding could pre-
vent the system from a total collapse. However, when to island and
how to ensure successful islanding, remains uncertain. The correct
identification and splitting of power system into various stable
islands is a challenging task that is currently being investigated
worldwide. Several algorithms have been developed for the identifi-
cation and splitting of power systems to stabilize islands. Few exam-
ples include the identification method based on the search
mechanism [15], comprehensive learning particle swarm optimiza-
tion (CLPSO) algorithm [17], ordered binary decision diagram (OBDD)
based algorithm [18], ANN based method [19], two stage stochastic
method [20], mixed integer linear programming (MILP) method
[21], angle modulated particle swarm optimization (AMPSO) and
slow coherency based method [22], algebraic graph based algorithm
[23], proper sizing and DG placement based algorithm [24], Line
 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152 141

Island Region 1

2 3 4 5 6 7 8
Island Region 3

DG DG

17

Utility Grid
9 14

22 18
23
10 15 DG 19
11
16 25 24
12 20

13 26
Island Region 2 21

Fig. 1. Islanding splitting strategy.

Micro Grid Test System For Observing the Islanding Effects Designed in the Different Countries of the world

United
States of
Asia Europe
America

Boston Bar − BC Hydro Canada

Japan China India Italy Netherlands United Greece CERTS Test Bed − United States
Germany
Kingdom
University of Wisconsin Micro Grid − United States

Boralex Planned Islanding − Hydro Quebec Canada

Kythnos Island Micro Grid − Greece

Laboratory Scale Micro Grid System at NTUA − Greece

DeMo Tec Test Micro Grid System − Germany

The Residential Micro Grid of Am Steinweg in Stutensee − Germany

University of Manchester Micro Grid Laboratory Prototype − UK

Bronsbergen Holiday Park Micro Grid − Netherland

CESI RICERCA DER Test Micro Grid − Italy

Test Micro Grid at IET − India

Laboratory Scale Micro Grid − China

Micro Grid Test Bed in Hefei University of Technology − China

Sendai Project − Japan

Hachinohe Project − Japan

Kyoto Eco Energy Project − Japan

Test Network at Akagi of CRIEPI − Japan

Aichi Micro Grid Project − Central Japan Airport City

Fig. 2. Micro Grid Test system designed in different countries of the world.

Table 1
Standards for islanding detection.

Standard Quality factor, Qf Islanding detection time, t Frequency range Voltage range
IEC 62116 1 t<2s (fo  1.5 Hz) 6 f and f 6 (fo + 1.5 Hz) 85% 6 V 6 115%
UL 1741 2.5 t<2s 59.3 Hz 6 f 6 60.5 Hz 88% 6 V 6 110%
IEEE 1547 1 t<2s 59.3 Hz 6 f 6 60.5 Hz 88% 6 V 6 110%
Korean Standard 1 t < 0.5 s 59.3 Hz 6 f 6 60.5 Hz 88% 6 V 6 110%
IEEE 929-2000 2.5 t<2s 59.3 Hz 6 f 6 60.5 Hz 88% 6 V 6 110%
VDE 0126-1-1 2 t < 0.2 s 47.5 Hz 6 f 6 50.5 Hz 88% 6 V 6 110%
142 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152
outage contingency based algorithm [25], Optimal intentional islan-
ding based on fast greedy algorithm and bloom algorithm [16].
After the islanding identification and splitting, system parame-
ters such as frequency and voltage should remain within the pre-
scribed limits. Otherwise, it may cause the islanded network to
experience blackouts. Under frequency load shedding (UFLS) tech-
nique is commonly applied to avoid this blackout [26]. Various
under frequency load shedding techniques have been proposed
to shed optimal loads in order to recover the frequency to nominal
value in islanding operation. Some of these UFLS schemes can be
found in [24,27–29]. Furthermore, intentional islanding of DG will
also influence electricity prices in the market. The effect of inten-
tional islanding on electricity prices has been examined in [30].
The paper recommended that some steps be taken to prevent elec-
tricity price spikes during islanding durations. This is due to the
fact that in the case of islanding, the whole power system will be
split into different islands, and each island will have its own price.
This may cause price spikes to the customers due to the DG own-
er’s monopoly. However, in cases where the generation is less than
the load within the island, the power may be supplied to customers
who are willing to pay, and disconnect the loads to those who are
not willing to pay [30]. The UFLS schemes that take into account
customers’ willingness to pay are discussed in [28,31].
Apart from these methods, various countries have installed
micro grid systems to evaluate the islanding effects and their
respective solutions. Fig. 2 presents the name and location of exist-
ing micro grid test networks around the world. The details of these
practical systems can be found in [32].
Moreover, several studies for the feasibility of islanding opera-
tion of different countries, such as United Kingdom [33,34], Caro-
lina [35], Thailand [36,37], India [38], Colombia [39], Brazil [40],
and Denmark [41] has been performed. The results show that the
operation of DGs in islanding mode has the potential to bring many
benefits to the distributed generator owner and customers. How-
ever, some of the technical studies have indicated the strong need
of reviewing their equipment and control strategies for successful
islanding operations [33,34].
3. Islanding detection standards and test frame
There are several international standards that can be used as
guide lines for utilities or independent power producers (IPP) to
design and operate the island system, such as IEEE Std. 1547-
2003 [3], IEEE Std. 929-2000 [42], UL 1741 [43], IEC 62116 [44],
and VDE 0126-1-1. Table 1 summarizes the standard parameters,
such as quality factor, islanding detection time, frequency, and
voltage operation range for these standards [45,46].
For a satisfactory operation of islanding detection technique,
IEEE 1547 standard has recommended a standard test system
PCC
DGBRK GBRK
DG Transformer
Grid
R L C more quickly as a result of the motor’s quicker regulation of power
Fig. 3. IEEE 1547 test frame for islanding detection.
frame as shown in Fig. 3. In this test frame, the islanding event
can be simulated by opening the grid circuit breaker (GBRK).
The type of load can affect the performance of islanding detec-
tion technique. According to IEEE 1547 test frame, such a load can
be modeled as a parallel RLC circuit. This circuit is primarily used,
as it raises more difficulties for the islanding detection techniques
compared to other combinations of circuit. The values of R, L, and C
for the unity power factor load can be determined using the follow-
ing equations:
R ¼ V 2 =P ð1Þ
L ¼ V 2 =2  p  f  Q f  P ð2Þ
C ¼ Q f  P=ð2  p  f  V 2 Þ ð3Þ
pffiffiffiffiffiffiffiffi
Q f ¼ R C=L ð4Þ
where Qf is the quality factor, which is the ratio of the amount of
energy stored in the load’s reactive elements to the amount of
energy dissipated in the load’s resistance. However, for non-unity
power factor, the inductance (L) and capacitance (C) can be calcu-
lated as such:
X L ¼ V 2 =Q ð5Þ
L ¼ X L =2  p  f ð6Þ
X C ¼ V 2 =Q ð7Þ
C ¼ 1=2  p  f  X C ð8Þ
Islanding detection technique commonly uses frequency, volt-
age, active power, and reactive power parameters as a mean to
detect the islanding event. The type of loads, affects all of these
parameters which in turn affect the performance of islanding
detection technique, as well as their non-detection zones. Several
researchers have conducted the research to observe the effects of
different types of loads (static load, dynamic load and composite
load) on the performance of islanding detection technique. Static
load model represents the active and reactive power of the load
as functions of voltage and frequency. Static load models can be
further divided into three types, namely constant power, constant
current, and constant impedance loads. Dynamic load models rep-
resent the induction motors loads, whereas composite loads com-
bine both static and dynamic type of loads to analyze their effect
on distribution system.
Zeineldin and Kirtley [47] have studied the impact of different
type’s static load on islanding detection. It has been observed that
over/under voltage and over/under frequency based methods
mostly depends on the frequency and voltage variations. Hence,
different types of loads can affect the voltage and frequency result-
ing in failure to detect islanding for large power mismatches, and
might create a very large non-detection zone [47]. Zhou et al.
[48] studied the impact of static load models and dynamic load
models on the frequency variations. The simulation results indi-
cated that among all these loads, the constant impedance load
model result in the worst frequency dynamics. Furthermore, it
was also observed that the inertia constant of an induction motor
also influenced the frequency dynamic curve. The results showed
that smaller inertia constant cause the frequency to be restored
compared to the larger inertia constant [48]. Another study on the
effects of static and composite loads on the voltage profile has been
presented in [49]. The study shows that voltage dips are highest for
composite loads, and lowest for static loads. Moreover, the study
has shown that during transient conditions, composite load are
 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152 143

Islanding
Detection
Techniques

Passive Active Hybrid Remote Computational

Islanding Islanding Islanding Islanding Intelligence Based
Detection Detection Detection Detection Islanding Detection
Techniques Techniques Techniques Techniques Techniques

Fig. 4. Types of Islanding Detection Techniques.

Receiver
Signal

Breaker
Grid
DG

Transmitter Load

Signal
DG
Load

Receiver

Fig. 5. Remote islanding detection scenario.

slow to respond to return to its original state [49]. Similarly, Abde-

laziz et al. [50] has presented the impact of static load and the Passive Islanding Detection Techniques
dynamic load on the distribution network. The simulation results
showed that the induction motor load and composite load model
reduce the voltage stability of the distribution system [50]. Thus,
it can be observed from the above discussion that different types
Change of Source Impedance Method

Rate of Change of Frequency Method

Harmonic Distortion Method

ROCOV & Change of Power Factor Method

ROCOF to Load Variation Method

Over/Under Voltage Method

Over/Under Frequency Method
Voltage Unbalance & THD of DG Current Method

of loads can affect the performance of islanding detection. Hence,

Voltage Unbalance Variation Method
ROCOF over Power Method

Rate of Change of Active Power Method

the islanding detection technique should be tested for all types

of loads for its reliable operation.

4. Islanding detection techniques and their types

Islanding detection techniques can be broadly divided into

remote and local islanding detection techniques. Local islanding
detections are further divided into passive and active islanding
detection methods [51–56]. Remote islanding detection techniques
rely on the communication between the utility and the DG site,
whereas, local techniques rely on the measurement of the system
parameter at the DG site. Fig. 4 shows the main classification of
islanding detection techniques.

4.1. Remote islanding detection techniques

Remote islanding detection techniques are based on the infor-
mation transferred through the communication between the util-
ity and the DG. Upon detecting the islanding event, the
communication signal is sent to trip DG units. Fig. 5 shows the
basic operating principle of remote islanding detection techniques.
It can be observed that the transmitter is placed at the
grid’s side, while the receiver is placed on each DG side, and
Fig. 6. Types of passive islanding detection techniques.
communication is carried out through a power line carrier. During
steady state operation when the circuit breaker is connected, the
receivers at the DG side continuously receive signals. However,
when the distribution network is islanded, the receiver will not
receive the signal due to the broken communication medium.
Hence, islanded is detected. This type of communication for
144 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152
islanding detection is known as power line signaling scheme [57–
59]. Another communication medium is the use of Supervisory
Control and Data Acquisition (SCADA) system for islanding detec-
tion. This type is known as the transfer trip scheme [60].
Remote islanding detection techniques have better reliability
than local techniques, however, these techniques are complex
and costly, especially for small distribution networks [61]. The
practical implementation of transfer trip scheme or power line sig-
naling scheme costs around $80,000–$250,000 for a single DG
installation [62]. Furthermore, these techniques highly rely only
on the communication means. Thus, any communication failure
may lead to malfunctions throughout the system. Due to these rea-
sons, remote techniques are not recommended for small DG sys-
tems [55]. Hence, local (passive and active) techniques are
widely used for islanding detection.
4.2. Passive islanding detection techniques
Passive islanding detection techniques use system parameters
measurement (voltage, frequency) at the DG terminals or point
of common coupling for islanding detection. These measurements
are compared with a predetermined threshold value for the pur-
pose of detecting islanding. Passive techniques are very cost effec-
tive, as they do not require large modifications in the protection
system [51–56]. Few common passive technique include the rate
of change of frequency [63–65], ratio of frequency variation to load
variation [66,67], rate of change of output power [60], rate of
change of frequency over power [66], voltage unbalance [68,69],
change of source impedance [70], harmonic distortion[68,71], volt-
age magnitude variation [72] and impedance based method to pro-
vide interlocking ROCOF relay [73]. Fig. 6 summarizes some
examples of passive islanding detection techniques.
Some passive islanding detection technique have also been pro-
posed using more than one parameter to detect islanding, such as
the rate of change of voltage and the change of power factor [74],
voltage unbalance, total harmonic distortion (THD) of DG current
[75], and frequency with damping factor of DG output frequency
[76]. The techniques proposed in [75,76] are suitable for inverter
and synchronous generator DGs. Another passive islanding detec-
tion technique suitable for single phase PV grid connected inverter
is proposed in [77]. The proposed technique was experimentally
developed in a low-cost PIC18F4550. The results showed that the
proposed technique detect islanding event effectively under vari-
ous load types, inverter output powers, and quality factors [77].
Other examples of passive techniques based on wavelet approach
can be found in [45,78,79].
The passive techniques have the advantage of not affecting the
power quality of the distribution network. Hence, power quality
issues, such as electrical noise, spikes, and voltage dip do not exist
Phase–PLL PErturbation Method
Negative Sequence Voltage Injection Method
High Frequency Signal Injection Method
Acve Islanding
Adaptive Logic Phase Shift Method
Techniques
Impedance Measurement Method
Reactive Power Export Error Detection Method
Current Injection Method
Fig. 7. Types of active islanding detection techniques.
in these techniques. However, passive techniques suffer from large
non-detection zone (NDZ), which is the range (in terms of power
difference between DG and load) where an islanding detection
technique fails to detect islanding. Furthermore, these techniques
need special care while setting the thresholds values. Setting up
of low threshold value may result in nuisance tripping, while high
threshold value setting may fail to detect islanding. The passive
technique drawbacks can be overcome by active islanding detec-
tion techniques.
4.3. Active islanding detection techniques
The principle of active techniques is based on the usage of high
frequency signals or some other means to slightly perturb the sys-
tem variables, such as voltage and frequency to detect the islan-
ding. The main concept behind active techniques is when the
distribution system is connected to the grid; the addition of pertur-
bation will cause a small variation in system parameter. However,
in the islanded mode, the system will observe a significant varia-
tion in system parameter, which will lead to the detection of islan-
ding. The examples of active islanding detection techniques
include slip mode frequency shift algorithms (SMFSA) [80], active
frequency drift (AFD), and active frequency drift with positive feed
back methods [81], automatic phase shift algorithm [82], sandia
frequency shift [83], sandia voltage shift [84], current injection
[85], negative sequence current injection [85–87], negative
sequence voltage injection [88], high frequency signal injection
[89–94], virtual capacitor [95], virtual inductor [96], phase –PLL
perturbation [97], adaptive logic phase shift method [98], method
based on voltage phase angle of inverter based DG [99] and
method based on magnitude changing of the injected current and
monitoring its voltage at the PCC [100]. Most of the active
techniques are employed for inverter-based distributed genera-
tions. However, only two techniques (impedance measurement
method [101] and reactive power export error detection method
[55,60]) were proposed for synchronous generator based DGs.
Fig. 7 summarizes some examples of active islanding detection
techniques.
Apart from this, most of the active islanding detection tech-
niques are generally proposed only for current controlled sources.
A technique suitable for voltage controlled sources has been pro-
posed in [102]. The proposed technique is verified experimentally,
and can detect islanding effectively [102]. Another active islanding
detection technique based on reference signal generator for grid
interconnected PV system is proposed in [103]. The proposed tech-
nique combines the reference signal generator with multi-level
inverter, which provides an efficient mean of detecting islanding.
The technique was verified experimentally, and is found capable
of detecting islanding within one half cycle of oscillation [103].
Negative Sequence Current Injection Method
Slip Mode Frequency Shift Algorithms
Detecon Virtual Capacitor / Virtual Inductor Method
Sandia Frequency Shift /Sandia Voltage Shift Method
Active Frequency Drift with Positive Feed Back Method
 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152 145

ROVC & Real Power Shift Method ANN 10

Hybrid Islanding Detection Techniques

Comp. Int. Based Techniques

Voltage Unbalance & Frequency Set Point Method FLC 7

Proportional Power Spectral Density & AFD Method ANFIS 5

Monitoring of HFC of DG Power, LCL Filter, Current Controller Method DT 10

Monitoring Current & Active Power of Circuit Breakers Method PSO 2

SMFS & Reactive Power versus Frequency Method AIS 1

ROCORP & Load Connecting Strategy Method SVM 2

0 2 4 6 8 10
Fig. 8. Types of hybrid islanding detection techniques.
Fig. 9. Computational intelligence based technique used for islanding detection.

Active islanding detection techniques have the advantage that

their non-detection zone is very small, and can detect islanding
even in perfect match of generation and load demand. However,
their main problem is that these techniques introduce perturbation
in the system at the regular intervals of time that are unnecessary
during most of the operating conditions, which often degrade the
quality of power. Furthermore, active technique takes large time
to detect islanding compared to passive techniques [104]. More-
over, most of the active techniques are limited to type of DG, such
as the change of terminal voltage may be more suitable for syn-
chronous generator based DG only due to their large reactive com-
ponent, whereas the frequency and phase shift methods are more
suitable for inverter type DG that only supplies more active power.
Meanwhile, active techniques work satisfactorily for single DG unit
only, and their response at multiple DG units is not guaranteed
[55]. However, efforts have been taken to find some suitable tech-
nique for multiple DG units, as discussed in [105,106]. Both works
has proposed an active technique based on the correlation tech-
nique for multiple DG, and compared it with the commonly used
voltage perturbation method. It has been concluded that for multi-
ple inverter based DGs, the correlation technique is effective in
detecting the islanding [105,106].
4.4. Hybrid islanding detection techniques
Hybrid islanding detection techniques employs the combined
features of active and passive techniques. In these techniques,
the active techniques are used only when passive technique is
unable to distinguish islanding events. The examples of hybrid
islanding detection techniques include methods based on voltage
unbalance and frequency set point [107], method based on rate
of voltage change and real power shift [4], method based on pro-
portional power spectral density and active frequency drift [108],
method based on monitoring of high frequency components of
DG output power, output LCL filter, current controller [109],
method based on rate of change of reactive power and load con-
necting strategy [110], method based on monitoring current and
active power of circuit breakers [111], and method based on slip
mode frequency shift and reactive power versus frequency [112].
Fig. 8 summarizes some examples of hybrid islanding detection
techniques.
These techniques have the advantage of acquiring a very small
NDZ. Furthermore, the injecting signals are not applied to the sys-
tem at every instant of time. Hence, the power degradation is
highly reduced. However, this combination increases the detection
time for islanding phenomena. Table 2 shows the comparison of
existing different islanding detection techniques.
From Table 2, it can be noticed that each techniques have their
respective strengths and limitations. However, the two main
important issue accuracy and fast detection are still unsolved by
these techniques. Hence, computational intelligence based tech-
nique for islanding detection may be a suitable option to accurately
detect the islanding phenomena within a short detection time.
5. Computational intelligence based techniques for islanding
detection
The term computational intelligent techniques generally refer
to a set of techniques that are applied to mimic human intelli-
gence. These techniques include Artificial Neural Networks
(ANN), Artificial Immune system (AIS), Fuzzy Logic Control (FLC),
Adaptive Neuro-fuzzy Inference System (ANFIS), Support Vector
Machine (SVM), and Decision Tree (DT) classifier for their opera-
tion. These techniques have the ability to solve those nonlinear
multi objective problems, which cannot be solved by the conven-
tional methods with desired speed and accuracy. The application
of these techniques for islanding detection has considerably grown.
Fig. 9 shows the usage of each technique for islanding detection,
which has been reviewed in this literature.
It can be noted from Fig. 9 that many types of intelligent tech-
niques have been implemented for islanding detection. From these
techniques, decision tree classifier and ANN have been utilized 10
times, followed by fuzzy logic control, 7 times. The implementa-
tions of each computational intelligence based technique for islan-
ding detection are presented in the following section.

Table 2
Comparison of existing islanding detection methods.

Parameter Remote technique Passive technique Active technique Hybrid technique

Principle Communication between DG and Monitoring changes in system Injecting frequency signals/applying Combination of passive and
utility system parameters disturbances active methods
Non detection None Large Small Very small
zone
Detection time Short Short Long Long
Power quality No degradation No degradation High degradation Very small degradation
Reliability High Low High High
Cost High Low Low Low
146 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152
5.1. Artificial Neural Network (ANN) based islanding detection
techniques
Artificial Neural Network (ANN) has been widely applied as a
common solution for variety of science and engineering problems
[113]. An ANN is a network of nodes or neurons analogous to the
biological synapse. Multi-layer feed forward networks are widely
adopted for power system problems. Various researchers have
applied ANN and its respective types for islanding detection appli-
cations. ANN based islanding detection techniques has been pro-
posed for multiple inverter based DG [114] and hybrid inverter
based DG [115]. The parameters employed for islanding detection
are transients in voltage signals [114] and transients signals in
three phase currents [115]; both techniques are categorized as pas-
sive islanding detection technique, which has no effect on power
quality. The technique used discrete wavelet transform to extract
the information for islanding detection. This extracted information
is then supplied to train ANN to differentiate between islanding
event and other disturbances. For training the ANN, various islan-
ding and non-islanding events are simulated, and discriminative
features of voltage signals at the DG side are captured. The non-
islanding cases include load increment and decrements, addition
and disconnection of capacitive load, single line to ground fault,
line-to-line fault, three phases to ground fault at different location
of distribution network, different DG operating states. The simula-
tion results shows the maximum accuracy of 97.77% [114] and
99.1% [115]. Furthermore, the technique proposed in [115] has
the ability to be implemented for real time applications.
Another ANN based hybrid islanding detection technique suit-
able for synchronous based DG is proposed in [116]. The technique
combines the passive and active technique to get better reliability
and accuracy. The passive technique has utilized six parameters
(Dv/dt, Df/dt, Df/Dp, Dp/dt, DQ/dt, and total harmonic distortion
(THD) of current), whereas an active technique uses the positive
feedback of active/reactive power. With the active power
approach, the variation of frequency is applied as an input to the
islanding detection compensator, and modifies the active power
reference to the DG, whereas in the reactive power approach, the
variation of voltage magnitude is applied as an input to islanding
detection compensator and modifies the reactive power reference
to the DG. This positive feedback further amplifies the voltage or
frequency departing from the limits, resulting in the detection of
islanding. It should be pointed out that this positive feedback has
no adverse effects during normal operation of the network. The
technique is tested on two distribution networks, and compared
with another technique proposed in [117] for the same network.
The simulation results showed that the technique has an accuracy
of 88.9%, whereas the accuracy of the technique proposed in [117]
was 83.33%. Furthermore, the technique used 6 parameters,
whereas the technique proposed in [117] uses 11 parameters (Df,
DV, Df/Dt, DV/Dt, Dp/Dt, Df/Dp, CTHD, VTHD, power factor devia-
tion, the absolute value of phase voltage times power factor, the
gradient of the voltage times power factor). Hence, the accuracy
of the detection algorithm is highly improved [116].
Another ANN based passive islanding detection technique suitable
for DFIG wind turbines based on the symmetrical components of the
second harmonic of voltage and current signals is proposed in [118].
The proposed technique used second order harmonic of voltage and
current measurements by processing these signals using Fourier
transform. The technique measures their symmetrical components
at the wind farm side and feed to an Artificial Neural Network
(ANN) for islanding detection. The results show that the proposed
ANN technique is able to distinguish between islanding and other
events, and can detect islanding in a very fast manner [118].
Apart from ANN, its types, such as self-organizing map (SOM)
neural network, extension neural network (ENN), probabilistic
neural network (PNN), and modular probabilistic neural network
(MPNN) have also been applied for islanding detection problems.
The application of SOM neural network for islanding detection is
proposed in [119,120]. The technique uses input signal of auto-
matic load frequency control to govern for islanding detection pur-
poses. SOM neural network is used to distinguish between
islanding and non-islanding events. The simulation results show
a maximum accuracy of 97.92% [119] and 98.19% [120],
respectively.
The application of ENN for islanding detection of PV-based DG is
proposed in [121]. The technique applied hybrid methods, where
the passive parameters selected are peak voltage, frequency, and
phase difference, whereas the active technique uses the voltage
drift method. The ENN was used to distinguish islanding event
and power quality disturbances (voltage swell, voltage dip, power
harmonic, and voltage flicker). The simulation results showed that
ENN based technique correctly detects islanding phenomena, and
is able to distinguish between power quality disturbances [121].
The application of PNN for islanding detection of multiple DG
was proposed in [122]. The technique is tested on a CIGRE medium
voltage distribution system using the PSCAD/EMTDC software. On
average, the overall classification obtained by PNN was 90%. The
application of MPNN for islanding detection of hybrid DG (fuel cell,
PV, and Wind) was proposed in [123]. The technique is proposed to
distinguish islanding and power quality disturbances. The tech-
nique applied S-transform to extract feature and used MPNN for
the classification. The technique is tested on experimental proto-
types with noise signals and different power quality disturbances.
The simulation results show that the proposed technique succeed
in obtaining 97.4% accuracy [123].
5.2. Fuzzy Logic Control (FLC) based islanding detection techniques
Fuzzy Logic Control (FLC) has emerged as a promising tool for
modeling a system that is not well defined by mathematical for-
mulation. FLC represent the expert human knowledge in the form
of linguistic variables called fuzzy rules. Fuzzy logic control has
also been applied for islanding detection problems. Fuzzy based
islanding detection techniques for rotating type DG has been pro-
posed in [124]. The paper uses three passive parameters (voltage,
df/dt, dp/dt) to detect islanding more accurately. The algorithm
monitors changes in three parameters and using fuzzy logic rules
to detect islanding events. The technique is verified on radial distri-
bution system with different types of load, and was found capable
of differentiating between islanding and non-islanding events
[124].
Fuzzy logic based technique considering 11 parameters (Df, DV,
Df/Dt, DV/Dt, Dp/Dt, Df/Dp, CTHD, VTHD, Dp.f, absolute (V/p.f), gra-
dient (V/p.f)) for islanding detection are proposed in [125,126]. The
papers perform initial classification by decision tree (DT), create
the fuzzy membership function from this DT classification, and
develop the corresponding rule base for islanding detection. The
technique is validated on system data, including both noise and
without noise. The simulation results showed that the fuzzy-based
islanding detection technique results in 100% accuracy. Further-
more, the online implementation of fuzzy logic is easier, which
makes it ideal for real time applications [125,126].
A combination of fuzzy rule based and S-transform application
for islanding detection is proposed in [127]. The technique used the
negative sequence voltage and negative sequence current as the
input parameters. The technique used fuzzy logic control to distin-
guish islanding and non-islanding event. The technique is vali-
dated with a large number of test cases, and shows very high
accuracies in correctly detecting islanding events. The detection
time was also very small (less than a cycle). The comparison of this
 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152 147

technique with other existing techniques showed that the pro-

posed technique has higher accuracy and fast detection [127]. Root
Another fuzzy based islanding detection technique suitable for
photovoltaic systems are based on active method of phase drift
control, is proposed in [128]. Active phase drift method is one of
the active detection methods, which controls the output current Node
of inverter to maintain a phase difference with the point of com-
mon coupling. When the grid disconnects, the PCC frequency devi-
ation will exceed the standard range, which leads to detection. The
simulation and experimental results show that the algorithm has a
Leaf Node
short detection time (6–20 cycles), and zero non detection zone
[128]. Aguiar et al. [129] has applied fuzzy logic control for islan-
ding detection using two techniques; voltage positive feedback,
and frequency positive feedback in dq synchronous frame. The pro-
posed technique is advantageous over other active techniques, in Leaf
that there is no injection disturbance at the PCC when the DG is
connected with the grid. Hence, islanding can be detected, and Fig. 10. Decision tree classifier basic structure.
power quality degradation is also highly reduced [129].
Sandia Frequency Shift (SFS) is one of the active islanding detec-
tion methods, which have very small NDZ. Due to this advantage, it integral controller with wavelet fuzzy neural network (WFNN)
is widely applied for inverter-based DG. However, NDZ in the SFS controller in order to improve the accuracy for islanding detection.
technique greatly rely on its design parameters. Improper tuning The technique is validated experimentally on a 2-kW current con-
of these parameters may result in the failure of this technique. trolled voltage source inverter according to UL1741 standard con-
To solve this problem, Vahedi and Karrari [130] proposed fuzzy figuration. Results show that the time taken with PI controller is
based technique, which estimates the load parameters (R, L, and 1.06 s, whereas the time taken with WFNN technique is 0.68 s.
C) online and adaptively tune the SFS parameter to eliminate Thus, the technique has very small islanding detection time, min-
NDZ in the SFS technique. The paper applied fuzzy logic control imum power quality degradation, and very small NDZs [132,133].
to estimate the amount of load parameters (R, L, and C) and conse- Hashemi et al. [134] proposed ANFIS based islanding detection
quently estimate the load quality factor. With this load quality fac- technique for inverter based DG. The technique used the rate of
tor estimation, a proper amount of positive feedback can be set in change of active power as input parameter, and applied it to ANFIS
the SFS formulation to prevent disability of islanding protection for distinguishing islanding event. This technique has the advan-
and nuisance tripping. The simulation results verify the excellent tage of detecting islanding independent of any threshold. The sim-
performance of the proposed method. Thus, the proposed tech- ulation results show that the NDZ of the proposed technique has
nique can be implemented in real time applications, because it been highly reduced to almost zero, and the problem of setting
can tune the SFS gain factor on a real-time basis [130]. thresholds is also eliminated [134]. Another application of ANFIS
in combination with discrete wavelet transform for islanding
5.3. Adaptive Neuro Fuzzy Inference System (ANFIS) Based Islanding detection of inverter based DG is presented in [135]. The technique
Detection Techniques is based on energy analysis of wavelet coefficients and adaptive
Neuro fuzzy inference system. Using the discrete wavelet theory,
ANFIS is a very powerful approach for modeling nonlinear and the energy of any decomposition level of all mother wavelet for
complex systems with less input and output training data. It com- parameters detection (df/dt, phase angle, d-q axis voltage, and
bines the advantages of ANN capability in learning from processes active/reactive power) is calculated, and the best signals are
and fuzzy logic control capability in handling uncertain informa-
tion. This makes ANFIS capable of approximating nonlinear and
uncertainties system without requiring pure mathematical mod-
els. ANFIS uses the Takagi–Sugeno fuzzy inference system. ANFIS Leaf Node Leaf Node
has also been applied for islanding detection problems. An Class 1 Class -1
ANFIS-based islanding detection technique was proposed in
[131]. The technique is based on a passive technique, using five Yes X4>d No
parameters (v, f, I, P, and f/p) and applied ANFIS to distinguish islan-
ding events. For this purpose, at the first stage, simulating the dis-
tribution system in PSCAD/EMTDC software for different scenarios Leaf Node Child Node Leaf Node Leaf Node
Class -1 X1<a, X2<b Class -1 Class 1
and measuring these five parameters results in massive amounts of
data being collected. At the second stage, this data is applied to
X2>b X3>c
ANFIS for training and testing to evaluate its effectiveness for islan- Yes No Yes No
ding detection. The results show that ANFIS based islanding detec-
tion technique results in 100% accuracy and consequently, zero
Child Node Child Node
NDZ. ANFIS-based islanding detection technique is easy to imple- X1>a X1<a
ment, quickly detects islanding event, and is applicable for multi-
ple as well as hybrid DGs. Thus, it can be applied in real time X1>a
Yes No
application, and for any types of DG [131].
Other ANFIS applications in combination with wavelet for islan-
ding detection are presented in [132,133]. Theses techniques are Root Node
consists of active technique, because they injects a current distur- (Including Entire data set)
bance signal, which deviate the frequency during an islanding
event. The technique replaces the conventional proportional Fig. 11. Decision tree with 3 child nodes and 5 leaf nodes.
148 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152
selected for ANFIS training for islanding detection purpose. Simula-
tion results show that the proposed algorithm reduces the NDZ
approximately to zero, and due to ANFIS, the requirement for set-
ting a threshold is eliminated [135].
5.4. Decision tree classifier based islanding detection techniques
Decision Tree (DT) classifier is considered as a type of pattern
recognition tool, which aims to provide a reasonable solution for
all possible inputs by taking into account their statistical varia-
tions. It is extremely useful for problems that cannot be solved
with analytical methods. The main advantage of DT is its fast train-
ing compared to other pattern recognition tools. The decision tree
algorithm is capable of breaking down complex decision-making
process into a union of several simpler decisions, making it easier
to interpret [136]. The basic DT structure is shown in Fig. 10 [122].
In the first step of the decision tree, the entire space is regarded
as a root node. An initial split is made using a predictor variable,
segmenting the root node into two child nodes. These child nodes
are chosen among all possible child nodes, and contain the purest
of data. Splits can be made from the child nodes. A leaf (terminal)
node is one where no more splits are being made [137]. Predictions
are made based on the make-up of leaf nodes. To use a decision
tree to make a prediction, the split decisions are followed until a
leaf node is reached. A simple 2-class decision tree with 3 child
nodes and 5 leaf nodes is shown in Fig. 11 [138].
The Decision tree classifier has been widely applied for islan-
ding detection techniques. A decision tree classifier based on the
islanding detection technique is proposed in [139]. The technique
proposes three different islanding detection methods based on
voltage, current, and both voltage, and current as a parameter,
and used discrete wavelet transform (DWT) to extract the features.
These features are processed to DT for islanding detection. The
technique is validated on a CIGRE medium distribution system.
The simulation results showed that among the three techniques,
the technique based on voltage signal has quick detection, simple
structure, and low cost compared to other techniques. This tech-
nique is accurate to up to 98%, and can detect islanding within
one cycle duration [139].
Lidula and Rajapakse [140] proposed a decision tree classifier
based islanding detection, utilizing voltage and current transient
signals to detect islanding events. The technique used DWT for fea-
ture extraction of these signals, and is used for training the deci-
sion tree classifier to distinguish islanding and non-islanding
events. The technique is validated on a CIGRE medium voltage dis-
tribution system with multiple DGs using PSCAD/EMTDC software.
The simulation results shows an accuracy of 96.43%, and can detect
islanding within two cycles duration [140]. The authors have pre-
sented the performance evaluation of this technique in another
paper, considered as part-II [141]. In this part, the technique is
applied to a test system, consisting of VSC based DC source. The
technique is compared with the other existing passive techniques
for multiple cases. The simulation results show that the DT based
technique has an accuracy of 96.11 ± 1.405%. The technique can
detect islanding within two cycles duration for synchronous gener-
ator based DG, and within three cycles duration for VSC based DC
source and the induction generator. Thus, the proposed technique
has higher accuracy, faster detection, and zero NDZ compared with
other passive techniques.
Pham et al. [142] investigated the hardware implementation of
the aforementioned technique, with the feature extraction part
replaced with analogue electronics to realize low cost. However,
with powerful digital signal processing (DSP) hardware available
today, fully digital implementation of the proposed algorithm is
not difficult. Furthermore, to test the superiority of decision tree
for classification, the same technique is compared with two other
pattern recognition techniques (support vector machine and mod-
ular probabilistic neural network) in [122]. On average, the overall
classification obtained by decision tree, MPNN, and support vector
machine was 99.61%, 90%, and 78%, respectively. The islanding
detection time of DT based technique was 0.0223 s, which is nearly
4 cycles faster than the ROCOF relay. Thus, DT-based technique
showed superior performance over MPNN and SVM based tech-
niques [122].
El-Arroudi et al. [117] proposed decision tree based islanding
detection technique, utilizing 11 parameters (Df, DV, Df/Dt, DV/
Dt, Dp/Dt, Df/Dp, CTHD, VTHD, power factor deviation, the abso-
lute value of phase voltage times power factor, the gradient of
the voltage times power factor) to detect islanding. The aim was
to propose a technique that can detect islanding for every network
topology. The information from these parameters was extracted
using the data mining approach, and is applied to DT based classi-
fier for islanding detection. The technique was tested on multiple
DG resources. The simulation results show that the technique
can successfully detect islanding event independent of thresholds
setting problem, and can detect within 45–50 ms duration [117].
Another application of decision tree for threshold setting of islan-
ding detection relays is proposed in [143]. It uses decision tree data
mining technology to extract threshold settings of islanding relays
from the analyses of system parameters (voltage, current, power,
power factor, frequency). The approach is tested on a multiple
DG resources, and the result indicates that this approach can be
used to optimize the detection threshold settings of the existing
islanding detection techniques. The advantage of using the pro-
posed approach is that it optimizes the threshold settings of islan-
ding relays, which consequently allow the detection of islanding
operation under minimum detectable zones for a system under
various operating conditions and for different network topologies
[143].
The decision tree classifier has also been combined with other
tools to enhance the accuracy for islanding detection. The applica-
tion of decision tree in combination with adaptive boosting (Ada-
Boost) technique for islanding detection is proposed in [138]. The
technique is suitable for PV, Doubly Fed Induction Generator
(DFIG) units, and synchronous generator based DG. Since the Deci-
sion tree algorithm cannot detect islanding in balanced condition,
hence, to increase the accuracy of classification and reduce the
NDZ region in such conditions, the AdaBoost algorithm is used.
AdaBoost algorithm linearly combines a set of weak classifiers in
order to generate a strong classifier. The simulation results showed
that the technique could detect islanding detection with negligible
non-detection zone [138]. Other applications of decision tree in
combination with fuzzy logic for islanding detection that took into
account 11 parameters are presented in [125,126]. It has been
shown in the results that the combination of fuzzy logic and deci-
sion tree succeed in obtaining 100% efficiency with and without
noise. Thus, it can be utilized for real time applications [125,126].
5.5. Application of other computational intelligence based techniques
for islanding detection
Apart from the above, few other computational intelligence
based techniques, such as artificial immune system (AIS) and sup-
port vector machine have also been used for islanding detection.
The AIS based islanding detection technique is proposed in [144].
The frequency of the inverter output voltage was used as the
detecting input signals and the algorithm was experimentally
implemented with DSP using a grid-connected inverter. Simulation
results show that the technique can accurately detect islanding
[144]. The application of support vector machine for islanding
detection is proposed in [122,123]. However, the accuracies
observed by both papers are quite different. The accuracy of the
 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152
Table 3
Percentage accuracies of Islanding Detection Techniques.
Islanding detection technique Accuracy (%)
Over/under voltage [141] 78.81
Over/under frequency [141] 90.24
Voltage vector shift (VVS) [141] 74.05
ROCOF based technique [141] 93.81
DT classifier based relay [140] 96.43
Intelligent based relay [117] 83.33
Voltage based DT relay [139] 98
ANN based technique [114] 97.77
ANN based technique [115] 99.1
ANN based technique [116] 88.9
technique proposed in [122] was 78%, whereas the accuracy of the
technique proposed in [123] was 98.2% [123], respectively.
6. Discussions
This paper has discussed the ability of computational intelli-
gence based technique in obtaining accurate islanding detection
within a short period of time. It can be noticed from the review that
the computational intelligence based techniques are the preferred
option for islanding detection application compared to passive,
active, and hybrid islanding detection methods. Computational
intelligent techniques have the ability to provide fast and possess
higher accuracies in islanding detection during grid disconnection.
Table 3 shows the accuracies of different passive, active, and intel-
ligent based techniques claimed by researchers.
It can be noticed from Table 3 that the computational intelli-
gence based techniques are mostly accurate to more than 95%,
few have efficiency up to or more than 99%, whereas ANFIS and
fuzzy with Decision tree based techniques claims 100% accuracy.
Furthermore, it can also be noted that computational intelligence
based techniques have also utilized the same parameters of pas-
sive, active, and hybrid islanding detection methods, but these
techniques have attained higher accuracy, due to their robustness
and dealing with complex system easily. Moreover, the successful
Islanding detection can lead to successful and viable islanding
operation of distributed generation in the near future to improve
the reliability and quality of power supply. Thus, computational
intelligence based techniques have the potential to be used for real
time applications to perform successful islanding operation of DGs.
7. Conclusion
This paper has presented a comprehensive review of computa-
tional intelligence based technique for islanding detection of dis-
tributed generations. The paper also provided review of existing
technique with their limitation requiring intelligent techniques
for higher accuracy and fast detection of islanding phenomenon.
The review has shown that compared with existing islanding
detection methods, the computational intelligence based tech-
nique has the highest accuracy (100%) and shorter islanding detec-
tion time. It can be concluded that the implementation of
computational intelligent techniques for islanding detection can
play an important role in performing successful and viable islan-
ding operation of DG. Their implementation in islanding detection
technique can also enhance the power system reliability and the
quality of the power supply. Thus, computational intelligence
based techniques can be preferable option for islanding detection
over existing islanding detection techniques.
149
Islanding detection Technique Accuracy (%)
SOM neural network [119] 97.92
SOM neural network [120] 98.19
PNN [122] 90
MPNN [123] 97.4
Fuzzy and DT based technique [125,126] 100
ANFIS based technique [131] 100
DT and DWT based technique [140] 96.43
DT based relay [122] 99.61
SVM based technique [122] 78
SVM based technique [123] 98.2
Acknowledgements
This work is supported by the Ministry of Education, Malaysia
under Research Acculturation Collaborative Effort (600-RMI/RACE
16/6/2, CR003-2014) and High Impact Research Grant (HIR-MOHE
D000004-16001), the University of Malaya and Quaid-e-Awam
University of Engineering Science & Technology Nawabshah, Sindh,
Pakistan.
References
[1] International Energy Agency. Distributed generation in liberalized electricity
market; 2002. <http://gasunie.eldoc.ub.rug.nl/FILES/root/2002/3125958/
3125958.pdf> (assessed 01.10.13).
[2] Laghari JA, MOKHLIS H, Bakar AHA, Karimi M, Mohammad H. Comparison of
fuzzy and PID based governor’s response in utilization of DG spinning reserve.
Przeglad Elektrotech. 2013;R. 89(NR 2a):187–90.
[3] IEEE Standard for Interconnecting Distributed Resources with Electric Power
Systems. IEEE Std 1547-2003; 2003. p. 1–16.
[4] Mahat P, Zhe C, Bak-Jensen B. A hybrid islanding detection technique using
average rate of voltage change and real power shift. IEEE Trans Power
Delivery 2009;24:764–71.
[5] Mohamad H, Mokhlis H, Bakar AHA, Ping HW. A review on islanding
operation and control for distribution network connected with small hydro
power plant. Renew Sustain Energy Rev 2011;15:3952–62.
[6] Mahat P, Zhe C, Bak-Jensen B. Review on islanding operation of distribution
system with distributed generation. IEEE power & energy soc gen meeting;
2011. p. 1–8.
[7] Basso TS, DeBlasio R. IEEE 1547 series of standards: interconnection issues.
IEEE Trans Power Electr 2004;19:1159–62.
[8] Hines P, Apt J, Talukdar S. Large blackouts in North America: historical trends
and policy implications. Energy Policy 2009;37:5249–59.
[9] Makarov YV, Reshetov VI, Stroev A, Voropai I. Blackout Prevention in the
United States, Europe, and Russia. IEEE Proc 2005;93:1942–55.
[10] Andersson G, Donalek P, Farmer R, Hatziargyriou N, Kamwa I, Kundur P, et al.
Causes of the 2003 major grid blackouts in North America and Europe, and
recommended means to improve system dynamic performance. IEEE Trans
Power Syst 2005;20:1922–8.
[11] Chang L, Wu Z. Performance and reliability of electrical power grids under
cascading failures. Int J Electr Power 2011;33:1410–9.
[12] Pourbeik P, Kundur PS, Taylor CW. The anatomy of a power grid blackout –
root causes and dynamics of recent major blackouts. IEEE Power Energy Mag
2006;4:22–9.
[13] Zhao X, Zhang X, He B. Study on self organized criticality of China power grid
blackouts. Energy Convers Manage 2009;50:658–61.
[14] Wang J-W, Rong L-L. Robustness of the western United States power grid
under edge attack strategies due to cascading failures. Safe Sci
2011;49:807–12.
[15] Aghamohammadi MR, Shahmohammadi A. Intentional islanding using a new
algorithm based on ant search mechanism. Int J Electr Power
2012;35:138–47.
[16] Pahwa S, Youssef M, Schumm P, Scoglio C, Schulz N. Optimal intentional
islanding to enhance the robustness of power grid networks. Physica A
2013;392:3741–54.
[17] El-Zonkoly A, Saad M, Khalil R. New algorithm based on CLPSO for controlled
islanding of distribution systems. Int J Electr Power 2013;45:391–403.
[18] Kai S, Da-Zhong Z, Qiang L. Splitting strategies for islanding operation of
large-scale power systems using OBDD-based methods. IEEE Trans Power
Syst 2003;18:912–23.
[19] Chen Y, Xu Z, Østergaard J. Security assessment for intentional island
operation in modern power system. Electr Power Syst Res 2011;81:1849–57.
[20] Golari M, Fan N, Wang J. Two-stage stochastic optimal islanding operations
under severe multiple contingencies in power grids. Electr Power Syst Res
2014;114:68–77.
150 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152
[21] Trodden PA, Bukhsh WA, Grothey A, McKinnon KIM. MILP formulation for
controlled islanding of power networks. Int J Electr Power 2013;45:501–8.
[22] Liu L, Liu W, Cartes DA, Chung I-Y. Slow coherency and Angle Modulated
Particle Swarm Optimization based islanding of large-scale power systems.
Adv Eng Inform 2009;23:45–56.
[23] Theodoro EAR, Benedito RAS, London Jr JBA, Alberto LFC. Algebraic-graph
method for identification of islanding in power system grids. Int J Electr
Power 2012;35:171–9.
[24] Shahmohammadi A, Ameli MT. Proper sizing and placement of distributed
power generation aids the intentional islanding process. Electr Power Syst
Res 2014;106:73–85.
[25] Hazarika D, Bhuyan S, Chowdhury SP. Line outage contingency analysis
including the system islanding scenario. Int J Electr Power 2006;28:
232–43.
[26] Vasquez-Arnez RL, Ramos DS, Del Carpio-Huayllas TE. Microgrid dynamic
response during the pre-planned and forced islanding processes involving
DFIG and synchronous generators. Int J Electr Power 2014;62:175–82.
[27] Ketabi A, Hajiakbari Fini M. An under frequency load shedding scheme for
islanded microgrids. Int J Electr Power 2014;62:599–607.
[28] Mokari-Bolhasan A, Seyedi H, Mohammadi-ivatloo B, Abapour S,
Ghasemzadeh S. Modified centralized ROCOF based load shedding scheme
in an islanded distribution network. Int J Electr Power 2014;62:806–15.
[29] Aghaebrahimi MR, Mehdizadeh M, Najafi HR. A new algorithm for reliability
assessment of wind-diesel system in islanding mode of operation. In: Int conf
on power eng energy and elect drives; 2011. p. 1–6.
[30] Zeineldin HH, Bhattacharya K, El-Saadany EF, Salama MMA. Impact of
intentional islanding of distributed generation on electricity market prices.
IET Gener Trans Distrib 2006;153:147–54.
[31] Mahat P, Zhe C, Bak-Jensen B. Underfrequency load shedding for an islanded
distribution system with distributed generators. IEEE Trans Power Deliver
2010;25:911–8.
[32] Lidula NWA, Rajapakse AD. Microgrids research: a review of experimental
microgrids and test systems. Renew Sustain Energy Rev 2011;15:186–202.
[33] Chowdhury SP, Chowdhury S, Crossley PA, Chui Fen T. UK scenario of islanded
operation of active distribution networks  A survey. IEEE power and energy
society general meeting- conversion and delivery of electrical energy in the
21st century; 2008. p. 1–6.
[34] Chowdhury SP, Chowdhury S, Crossley PA. UK scenario of islanded operation
of active distribution networks with renewable distributed generators. Int J
Electr Power 2011;33:1251–5.
[35] Gooding PA, Makram E, Hadidi R. Probability analysis of distributed
generation for island scenarios utilizing Carolinas data. Electr Power Syst
Res 2014;107:125–32.
[36] Fuangfoo P, Wei-Jen L, Ming-Tse K. Impact study on intentional islanding of
distributed generation connected to a radial subtransmission system in
Thailand’s electric power system. IEEE Trans Ind Appl 2007;43:1491–8.
[37] Fuangfoo P, Wei-Jen L, Nigim KA. Intentional islanding operation to improve
the service reliability of Thailand electric power system. IEEE power
engineering society general meeting; 2007. p. 1–7.
[38] Joshi KA, Pindoriya NM. Risk assessment of unintentional islanding in a spot
network with roof-top photovoltaic system  a case study in India. IEEE
innovative smart grid tech – Asia; 2013. p. 1–6.
[39] Carvajal Quintero SX, Marin Jimenez JD, Arango Aramburo S. Feasibility of
intentional islanding operation with small hydropower plants. In: Sixth IEEE
Latin America conference and exposition on transmission and distribution;
2012. p. 1–6.
[40] Londero RR, Affonso CM, Nunes MVA, Freitas W. Planned islanding for
Brazilian system reliability. In: IEEE conference and exposition on
transmission and distribution; 2010. p. 1–6.
[41] Yu C, Zhao X, Ostergaard J. Frequency analysis for planned islanding operation
in the Danish distribution system – Bornholm. In: 43rd Int universities power
engineering conference; 2008. p. 1–5.
[42] IEEE Recommended Practice for Utility Interface of Photovoltaic (PV)
Systems. IEEE Std 929-2000; 2000.
[43] UL1741. Inverter, converter, and controllers for use in independent power
system; 2001.
[44] IEC 62116 Edition 1.0. International standard, test procedure of islanding
prevention measures for utility-interconnected photovoltaic inverters; 2008–
09.
[45] Ku Ahmad KNE, Selvaraj J, Rahim NA. A review of the islanding detection
methods in grid-connected PV inverters. Renew Sustain Energy Rev
2013;21:756–66.
[46] Yu B, Matsui M, Yu G. A review of current anti-islanding methods for
photovoltaic power system. Sol Energy 2010;84:745–54.
[47] Zeineldin HH, Kirtley JL Jr. Islanding operation of inverter based Distributed
Generation with static load models. IEEE power and energy society general
meeting – conversion and delivery of electrical energy in the 21st century;
2008. p. 1–6.
[48] Zhou B, Han M, Tang X, Wu Y. Electric load dynamic frequency modelling and
its application for islanding operation control. In: Int conf on sustainable
power generation and supply; 2012. p. 1–6.
[49] Roy NK, Hossain MJ, Pota HR. Effects of load modeling in power distribution
system with distributed wind generation. In: Universities power engineering
conference; 2011. p. 1–6.
[50] Abdelaziz S, Khadija BK, Mohamed E. Assessment of wind power penetration
level in distribution network with consideration of static, motor and
composite loads. In: 5th International renewable energy congress (IREC);
2014. p. 1–6.
[51] Funabashi T, Koyanagi K, Yokoyama R. A review of islanding detection
methods for distributed resources. In: IEEE proc power tech conference;
2003. p. 6.
[52] Kunte RS, Wenzhong G. Comparison and review of islanding detection
techniques for distributed energy resources. In: IEEE North American power
symposium, Calgary, Canada; 2008. p. 1–8.
[53] Mahat P, Zhe C, Bak-Jensen B. Review of islanding detection methods for
distributed generation. Electric utility deregulation and restructuring and
power technologies; 2008. p. 2743–8.
[54] Velasco D, Trujillo CL, Garcerá G, Figueres E. Review of anti-islanding
techniques in distributed generators. Renew Sustain Energy Rev
2010;14:1608–14.
[55] Jun Y, Liuchen C, Diduch C. Recent developments in islanding detection for
distributed power generation. In: Large engineering systems conference on
power engineering; 2004. p. 124–8.
[56] Menon V, Nehrir MH. A review of issues regarding the use of distributed
generators. In: Proceedings of the 37th annual North American power
symposium; 2005. p. 399–405.
[57] Ropp ME, Aaker K, Haigh J, Sabbah N. Using power line carrier
communications to prevent islanding [of PV power systems]. In: IEEE
photovoltaic specialists conference; 2000. p. 1675–8.
[58] Wencong W, Kliber J, Guibin Z, Wilsun X, Howell B, Palladino T. A power line
signaling based scheme for anti-islanding protection of distributed
generators Part II: Field test results. IEEE Trans Power Deliver
2007;22:1767–72.
[59] Wilsun X, Guibin Z, Chun L, Wencong W, Guangzhu W, Kliber J. A power line
signaling based technique for anti-islanding protection of distributed
generators—Part I: Scheme and analysis. IEEE Trans Power Deliver
2007;22:1758–66.
[60] Redfern MA, Usta O, Fielding G. Protection against loss of utility grid supply
for a dispersed storage and generation unit. IEEE Trans Power Deliver
1993;8:948–54.
[61] Barsali S, Ceraolo M, Pelacchi P, Poli D. Control techniques of Dispersed
Generators to improve the continuity of electricity supply. In: IEEE power
engineering society winter meeting, vol. 2; 2002. p. 789–94.
[62] Wrinch Michael C. Negative sequence impedance measurement for distributed
generator islanding detection. Electrical and computer engineering, faculty of
graduate studies. The University of British Columbia; 2008.
[63] Warin J, Allen WH. Loss of mains protection. In: Proc ERA conf circuit
protection for industrial and commercial installation, London, UK; 1990. pp.
4.3.1–4.3.12.
[64] Affonso CM, Freitas W, Xu W, Da Silva LCP. Performance of ROCOF relays for
embedded generation applications. IET Gener Trans Distrib
2005;152:109–14.
[65] Freitas W, Wilsun X, Affonso CM, Zhenyu H. Comparative analysis between
ROCOF and vector surge relays for distributed generation applications. IEEE
Trans Power Deliver 2005;20:1315–24.
[66] Fu-Sheng P, Shyh-Jier H. A detection algorithm for islanding-prevention of
dispersed consumer-owned storage and generating units. IEEE Trans Energy
Convers 2001;16:346–51.
[67] Pai FS, Huang SJ. A detection algorithm for islanding-prevention of dispersed
consumer-owned storage and generating units. IEEE Power Eng Rev
2001;21:67.
[68] Jang SI, Kim KH. A new islanding detection algorithm for distributed
generations interconnected with utility networks. In: IEEE int conf on
developments in power system protection; 2004. p. 571-4.
[69] Kamyab Ebadollah, Sadeh Javad. Islanding detection method for photovoltaic
distributed generation based on voltage drifting. IET Gener Trans Distrib
2013:584–92.
[70] Hopewell PD, Jenkins N, Cross AD. Loss-of-mains detection for small
generators. In: IEE proceedings on electric power applications, vol. 143;
1996. p. 225-30.
[71] Kobayashi H, Takigawa K, Hashimoto E, Kitamura A, Matsuda H. Method for
preventing islanding phenomenon on utility grid with a number of small
scale PV systems. In: IEEE photovoltaic specialists conference; 1991. p. 695–
700.
[72] Kim JE, Hwang JS. Islanding detection method of distributed generation units
connected to power distribution system. In: Int conf on power system
technology, vol. 2; 2000. p. 643–7.
[73] Jia K, Bi T, Liu B, Thomas D, Goodman A. Advanced islanding detection utilized
in distribution systems with DFIG. Int J Electr Power 2014;63:113–23.
[74] Salman SK, King DJ, Weller G. New loss of mains detection algorithm for
embedded generation using rate of change of voltage and changes in power
factors. In: IEEE int conf on developments in power system protection; 2001.
p. 82–5.
[75] Sung-Il J, Kwang-Ho K. Development of a logical rule-based islanding
detection method for distributed resources. In: IEEE power engineering
society winter meeting; 2002. p. 800–6.
[76] Zeineldin HH, Abdel-Galil T, El-Saadany EF, Salama MMA. Islanding detection
of grid connected distributed generators using TLS-ESPRIT. Electr Power Syst
Res 2007;77:155–62.
[77] Ahmad KNEK, Rahim NA, Selvaraj J, Rivai A, Chaniago K. An effective passive
islanding detection method for PV single-phase grid-connected inverter. Sol
Energy 2013;97:155–67.
 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152
[78] Hsieh C-T, Lin J-M, Huang S-J. Enhancement of islanding-detection of
distributed generation systems via wavelet transform-based approaches.
Int J Electr Power 2008;30:575–80.
[79] Lidula NWA, Perera N, Rajapakse AD. Investigation of a fast islanding
detection methodology using transient signals. In: IEEE power & energy
society general meeting; 2009. p. 1–6.
[80] Smith GA, Onions PA, Infield DG. Predicting islanding operation of grid
connected PV inverters. In: IEE proceedings on electric power applications,
vol. 147; 2000. p. 1–6.
[81] Ropp ME, Begovic M, Rohatgi A. Analysis and performance assessment of the
active frequency drift method of islanding prevention. IEEE Trans Energy
Convers 1999;14:810–6.
[82] Guo-Kiang H, Chih-Chang C, Chen C-L. Automatic phase-shift method for
islanding detection of grid-connected photovoltaic inverters. IEEE Trans
Energy Convers 2003;18:169–73.
[83] Zeineldin HH, Salama MMA. Impact of load frequency dependence on the
NDZ and performance of the SFS islanding detection method. IEEE Trans Ind
Electron 2011;58:139–46.
[84] Stevens J, Bonn R, Ginn J, Gonzalez S, Kern G. Development and testing of an
approach to anti-islanding in utility-interconnected photovoltaic systems.
Sandia National Laboratories report SAND2000-1939, Sandia National
Laboratories, Albuquerque, NM, August; 2000.
[85] Hernandez-Gonzalez G, Iravani R. Current injection for active islanding
detection of electronically-interfaced distributed resources. IEEE Trans Power
Deliver 2006;21:1698–705.
[86] Karimi H, Yazdani A, Iravani R. Negative-sequence current injection for fast
islanding detection of a distributed resource unit. IEEE Trans Power Electr
2008;23:298–307.
[87] Tuyen ND, Fujita G. Negative-sequence current injection of dispersed
generation for islanding detection and unbalanced fault ride-through. In:
Int conf universities’ power engineering; 2011. p. 1–6.
[88] Byeong-Heon K, Seung-Ki S, Chun-Ho L. Anti-islanding detection method
using Negative Sequence Voltage. In: Int conf power electronics and motion
control; 2012. p. 604–8.
[89] Briz F, Reigosa D, Blanco C, Guerrero J. Coordinated operation of parallel-
connected inverters for active islanding detection using high frequency signal
injection. IEEE Trans Ind Appl 2014. http://dx.doi.org/10.1109/
tia.2014.2305802.
[90] Briz F, Reigosa D, Blanco C, Guerrero JM. Coordinated operation of parallel-
connected inverters for active islanding detection using high frequency signal
injection. In: IEEE energy conversion congress and exposition; 2013. p. 2296–
303.
[91] Reigosa D, Briz F, Blanco C, Garcia P, Guerrero JM. Active islanding detection
using high frequency signal injection. In: IEEE energy conversion congress
and exposition; 2011. p. 2183–90.
[92] Reigosa D, Briz F, Blanco C, Garcia P, Guerrero JM. Active islanding detection
for multiple parallel- connected inverter-based distributed generators using
high frequency signal injection. In: IEEE energy conversion congress and
exposition; 2012. p. 2719–26.
[93] Reigosa D, Briz F, Blanco C, Garcia P, Manuel Guerrero J. Active islanding
detection for multiple parallel-connected inverter-based distributed
generators using high-frequency signal injection. IEEE Trans Power Electr
2014;29:1192–9.
[94] Reigosa DD, Briz F, Charro CB, Garcia P, Guerrero JM. Active islanding
detection using high-frequency signal injection. IEEE Trans Ind Appl
2012;48:1588–97.
[95] Chiang W-J, Jou H-L, Wu J-C. Active islanding detection method for inverter-
based distribution generation power system. Int J Electr Power
2012;42:158–66.
[96] Jou HL, Chiang WJ, Wu JC. Virtual inductor-based islanding detection method
for grid-connected power inverter of distributed power generation system.
IET Renew Power Gener 2007;1:175–81.
[97] Velasco D, Trujillo C, Garcera G, Figueres E. An active anti-islanding method
based on phase-PLL perturbation. IEEE Trans Power Electr 2011;26:
1056–66.
[98] Jun Y, Liuchen C, Diduch C. A new adaptive logic phase-shift algorithm for
anti-islanding protections in inverter-based DG systems. In: IEEE power
electronics specialists conference; 2005. p. 2482–6.
[99] Pourbabak H, Kazemi A. A new technique for islanding detection using
voltage phase angle of inverter-based DGs. Int J Electr Power
2014;57:198–205.
[100] Indu Rani B, Srikanth M, Saravana Ilango G, Nagamani C. An active islanding
detection technique for current controlled inverter. Renew Energy
2013;51:189–96.
[101] O’Kane P, Fox B. Loss of mains detection for embedded generation by system
impedance monitoring. In: Int conf on developments in power system
protection; 1997. p. 95–8.
[102] Shen J-M, Jou H-L, Wu J-C, Wu K-D. Hybrid photovoltaic generation system
with novel islanding detection method. Electr Power Syst Res
2014;106:101–8.
[103] Tsang KM, Chan WL. Rapid islanding detection using multi-level inverter for
grid-interactive PV system. Energy Convers Manage 2014;77:278–86.
[104] Mahat P, Zhe C, Bak-Jensen B. Review of islanding detection methods for
distributed generation. In: Int conf on electric utility deregulation and
restructuring and power technologies; 2008. p. 2743–8.
151
[105] Tsukamoto O, Yamagishi K. Detection of islanding of multiple dispersed
photovoltaic power systems. Sol Energy 1996;58:9–15.
[106] Tsukamoto O, Okayasu T, Yamagishi K. Study on islanding of dispersed
photovoltaic power systems connected to a utility power grid. Sol Energy
2001;70:505–11.
[107] Menon V, Nehrir MH. A hybrid islanding detection technique using voltage
unbalance and frequency set point. IEEE Trans Power Syst 2007;
22:442–8.
[108] Jun Y, Diduch CP, Liucheng C. Islanding detection using proportional power
spectral density. IEEE Trans Power Deliver 2008;23:776–84.
[109] Pigazo A, Moreno VM, Liserre M, Aquila AD. Wavelet-Based islanding
detection algorithm for single-phase photovoltaic (PV) distributed
generation systems. In: IEEE int symposium on industrial electronics; 2007.
p. 2409–13.
[110] Laghari JA, Mokhlis H, Bakar AHA, Karimi M. A new islanding detection
technique for multiple mini hydro based on rate of change of reactive power
and load connecting strategy. Energy Convers Manage 2013;76:
215–24.
[111] Bayrak G, Cebeci M. A novel anti islanding detection method for grid
connected fuel cell power generation systems. Int J Hydrogen Energy
2014;39:8872–80.
[112] Akhlaghi S, Ghadimi AA, Akhlaghi A. A novel hybrid islanding detection
method combination of SMS and Q-f for islanding detection of inverter-
based DG. In: Power and energy conference at Illinois; 2014. p. 1–8.
[113] Dias FM, Antunes A, Mota AM. Artificial neural networks: a review of
commercial hardware. Eng Appl Artif Intel 2004;17:945–52.
[114] Fayyad Y, Osman A. Neuro-wavelet based islanding detection technique. In:
IEEE electric power and energy conference; 2010. p. 1–6.
[115] ElNozahy MS, El-Saadany EF, Salama MMA. A robust wavelet-ANN based
technique for islanding detection. In: IEEE power and energy society general
meeting; 2011. p. 1–8.
[116] Ghazi R, Lotfi N. A new hybrid intelligent based approach to islanding
detection in distributed generation. In: Int universities power engineering
conference; 2010. p. 1–5.
[117] El-Arroudi K, Joos G, Kamwa I, McGillis DT. Intelligent-based approach to
islanding detection in distributed generation. IEEE Trans Power Deliver
2007;22:828–35.
[118] Abd-Elkader AG, Allam DF, Tageldin E. Islanding detection method for DFIG
wind turbines using artificial neural networks. Int J Electr Power
2014;62:335–43.
[119] Moeini A, Darabi A, Karimi M. Clustering governor signal of distributed
generation for islanding detection. In: IEEE conf on computational tech in
electrical and electronics engineering; 2010. p. 493–8.
[120] Moeini A, Darabi A, Rafiei SMR, Karimi M. Intelligent islanding detection of a
synchronous distributed generation using governor signal clustering. Electr
Power Syst Res 2011;81:608–16.
[121] Chao K-H, Chiu C-L, Li C-J, Chang Y-C. A novel neural network with simple
learning algorithm for islanding phenomenon detection of photovoltaic
systems. Exp Syst Appl 2011;38:12107–15.
[122] Lidula NWA, Rajapakse AD. Fast and reliable detection of power islands using
transient signals. In: Int conf on industrial and information systems; 2009. p.
493–8.
[123] Mohanty SR, Ray PK, Kishor N, Panigrahi BK. Classification of disturbances in
hybrid DG system using modular PNN and SVM. Int J Electr Power
2013;44:764–77.
[124] Rosolowski Eugeniusz, Arkadiusz Burek, Leszek Jedut. A new method for
islanding detection in distributed generation. Wroclaw University of
Technology, Department of Electrical Power Engineering, Poland; 2007.
[125] Samantaray SR, El-Arroudi K, Joos G, Kamwa I. A fuzzy rule-based approach
for islanding detection in distributed generation. IEEE Trans Power Deliver
2010;25:1427–33.
[126] Kumarswamy I, Sandipamu TK, Prasanth V. Analysis of islanding detection in
distributed generation using fuzzy logic technique. 7th Asia modelling
symposium; 2013. p. 3–7.
[127] Dash PK, Padhee M, Panigrahi TK. A hybrid time–frequency approach based
fuzzy logic system for power island detection in grid connected distributed
generation. Int J Electr Power 2012;42:453–64.
[128] Liang S, Fangde W. An islanding detection algorithm based on fuzzy adaptive
phase drift control. In: IEEE int conf on information and automation; 2013. p.
225–9.
[129] Aguiar CR, Bastos RF, Neves RVA, Reis GB, Machado RQ. Fuzzy positive
feedback for islanding mode detection in distributed generation. In: IEEE
power and energy society general meeting; 2013. p. 1–5.
[130] Vahedi H, Karrari M. Adaptive fuzzy sandia frequency-shift method for
islanding protection of inverter-based distributed generation. IEEE Trans
Power Deliver 2013;28:84–92.
[131] Bitaraf H, Sheikholeslamzadeh M, Ranjbar AM, Mozafari B. Neuro-fuzzy
islanding detection in distributed generation. IEEE innovative smart grid
technologies; 2012. p. 1–5.
[132] Faa-Jeng L, Kuang-Hsiung T, Jian-Hsing C. Active islanding detection method
using wavelet fuzzy neural network. In: IEEE int conf on fuzzy systems; 2012.
p. 1–8.
[133] Faa-Jeng L, Yi-Sheng H, Kuang-Hsiung T, Jian-Hsing C, Yung-Ruei C. Active
islanding detection method using d-axis disturbance signal injection with
intelligent control. IET Gener Trans Distrib 2013;7:537–50.
152 J.A. Laghari et al. / Energy Conversion and Management 88 (2014) 139–152
[134] Hashemi F, Ghadimi N, Sobhani B. Islanding detection for inverter-based DG
coupled with using an adaptive neuro-fuzzy inference system. Int J Electr
Power 2013;45:443–55.
[135] Shayeghi H, Sobhani B. Zero NDZ assessment for anti-islanding protection
using wavelet analysis and neuro-fuzzy system in inverter based distributed
generation. Energy Convers Manage 2014;79:616–25.
[136] Yong S, Rovnyak SM. Decision tree-based methodology for high impedance
fault detection. IEEE Trans Power Deliver 2004;19:533–6.
[137] Rokach L, Maimon O. Top-down induction of decision trees classifiers - a
survey. IEEE Trans Syst Man Cybernetics Part C Appl Rev 2005;
35:476–87.
[138] Madani SS, Abbaspour A, Beiraghi M, Dehkordi PZ, Ranjbar AM. Islanding
detection for PV and DFIG using decision tree and AdaBoost algorithm. In:
IEEE int conf and exhibition on innovative smart grid technologies; 2012. p.
1–8.
[139] Heidari M, Seifossadat G, Razaz M. Application of decision tree and discrete
wavelet transform for an optimized intelligent-based islanding detection
method in distributed systems with distributed generations. Renew Sustain
Energy Rev 2013;27:525–32.
[140] Lidula NWA, Rajapakse AD. A pattern recognition approach for detecting
power islands using transient signals—Part I: Design and implementation.
IEEE Trans Power Deliver 2010;25:3070–7.
[141] Lidula NWA, Rajapakse AD. A pattern-recognition approach for detecting
power islands using transient signals—Part II: performance evaluation. IEEE
Trans Power Deliver 2012;27:1071–80.
[142] Pham JP, Denboer N, Lidula NWA, Perera N, Rajapakse AD. Hardware
implementation of an islanding detection approach based on current and
voltage transients. In: IEEE electrical power and energy conference; 2011. p.
152–7.
[143] El-Arroudi K, Joos G. Data mining approach to threshold settings of islanding
relays in distributed generation. IEEE Trans Power Syst 2007;22:1112–9.
[144] Guiliang Y. A distributed generation islanding detection method based on
artificial immune system. In: IEEE transmission and distribution conf and
exhibition: Asia and Pacific; 2005. p. 1–4.