Energy Conversion and Management 96 (2015) 613–624

Contents lists available at ScienceDirect

Energy Conversion and Management

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

Application of signal processing techniques for islanding detection of

distributed generation in distribution network: A review
Safdar Raza a, Hazlie Mokhlis a,⇑, Hamzah Arof a, J.A. Laghari a, Li Wang b
a
Department of Electrical Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia
b
Department of Electrical Engineering, College of Electrical Engineering & Computer Science, National Cheng Kung University, Tainan City 70101, Taiwan

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

Article history: High penetration of distributed generation resources (DGR) in distribution network provides many bene-
Received 5 December 2014 fits in terms of high power quality, efficiency, and low carbon emissions in power system. However, effi-
Accepted 6 March 2015 cient islanding detection and immediate disconnection of DGR is critical in order to avoid equipment
damage, grid protection interference, and personnel safety hazards. Islanding detection techniques are
mainly classified into remote, passive, active, and hybrid techniques. From these, passive techniques
Keywords: are more advantageous due to lower power quality degradation, lower cost, and widespread usage by
Islanding detection
power utilities. However, the main limitations of these techniques are that they possess a large non
Signal processing
Fourier transform
detection zones and require threshold setting. Various signal processing techniques and intelligent clas-
Wavelet transform sifiers have been used to overcome the limitations of passive islanding. Signal processing techniques, in
s-transform particular, are adopted due to their versatility, stability, cost effectiveness, and ease of modification. This
Hilbert Huang transform paper presents a comprehensive overview of signal processing techniques used to improve common pas-
tt-transform sive islanding detection techniques. A performance comparison between the signal processing based
islanding detection techniques with existing techniques are also provided. Finally, this paper outlines
the relative advantages and limitations of the signal processing techniques in order to provide basic
guidelines for researchers and field engineers in determining the best method for their system.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction phenomenon of a distribution network connected with three dis-

Due to the rapid increase of oil and natural gas prices, green-
house effect, and other environmental issues, the installation of
distributed generation resources (DGRs) has shed a new light on
the field of electric power supply. It attracts industrial and com-
mercial customers by providing a competitive environment with
respect to new integrating technologies, environmental benefits,
and reduced losses [1,2]. The DGRs, such as mini/micro hydro,
wind turbines, photovoltaic and fuel cell increases the efficiency
and stability of the distribution network [3]. However, the steady
state and dynamic behavior of distributed energy resources causes
technical challenges in control and protection [4]. The occurrence
of islanding is one of the major issues in this context. In islanding
condition, the part of the distribution system to which DGR is con-
nected is electrically isolated from the utility grid but energized by
the distributed generation resource [5]. Fig. 1 shows the islanding
⇑ Corresponding author. Tel.: +60 3 79675238 (office), mobile: +60 1 26467936;
fax: +60 03 79675316.
E-mail address: hazlie@um.edu.my (H. Mokhlis).
tribution generators (DGs).
If the utility supply is cut off, then an island is formed, as shown
by dotted circles. Islanding creates many problems, such as com-
plication in orderly power restoration, degradation of power qual-
ity, and endangering the lives of utility personnel. Due to these
severe consequences, existing standards such as IEC 62116, IEEE
1547, IEEE 929-2000 and AS4777.3-2005 standards do not allow
distributed generation resources to operate in islanded mode [6].
According to IEEE 1547 standard, the islanding condition should
be detected and disconnected within 2 s from the distribution net-
work [7].
According to aforementioned standards, the distributed genera-
tion resources are disconnected from the power system when
islanding occurs. However, it is not an economical solution,
because power generation from DGRs may fulfil the load demand,
if not the whole demand, as part of its operation. In this regard,
researchers are working expansively in order to make the islanding
operation of distribution network a viable solution by properly
addressing various technical issues [8].
The most important and the foremost step for successful island-
ing are to detect the islanding condition when the distribution

http://dx.doi.org/10.1016/j.enconman.2015.03.029
0196-8904/Ó 2015 Elsevier Ltd. All rights reserved.
614 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624
Fig. 1. Phenomenon of islanding occurrence in distribution network connected
with three DG.
network is disconnected from the grid. It requires an efficient
islanding detection technique in order to detect and to isolate
the distributed energy resource. If the islanding condition is not
detected accurately and timely, the whole distribution network
might collapse. Thus, accuracy and time are the two most impor-
tant factors in islanding detection. Several techniques have been
proposed for this, and they are categorized as remote and local
(passive, active, and hybrid) methods. However, each method is
replete with its own merits and demerits [9]. Hence, the research-
ers shift their interests and intentions towards signal processing
based islanding detection techniques, because these techniques
have the ability to extract the unseen/hidden features of the mea-
sured signal that are helpful for islanding detection. The objective
of this paper is to review the existing signal processing based
islanding detection techniques.
The rest of the paper is organized in the following order.
Section 2 elaborates the islanding detection techniques. Signal pro-
cessing based islanding detection techniques, with and without
intelligent classifiers, is covered under Sections 3 and 4. Some mis-
cellaneous techniques and discussion is presented in Sections 5
and 6. Finally, the conclusion is presented in Section 7.
2. Categorization of islanding detection methods
Islanding detection techniques are mainly categorized into
remote, local, and signal processing based, as shown in Fig 2
[10–12]. These methods are further classified into different
techniques on the basis of different parameters, such as detec-
tion speed, error detection rate, power quality, non detection zone
(NDZ), and efficacy in multiple inverter cases [13]. Comprehensive
discussions of these techniques are presented in the following
sections.
2.1. Remote islanding detection techniques
Remote islanding detection techniques works on the principle
of communication between utility and distributed energy resource.
Once islanding occurs, a trip signal is sent to the distributed
generation resource. Transfer trip scheme [14,15] and Power line
carrier communication (PLCC) scheme [16,17] falls under this cat-
egory. These techniques have zero NDZ, faster response time, zero
impact on power quality and system transients, high reliability,
and works effectively in multiple DG systems. However, remote
techniques are very expensive for implementation on small-scale
systems [18]. Hence, local techniques are preferred for the afore-
mentioned purpose.
2.2. Local islanding detection techniques
As the name indicates, it is based on measuring the variations in
system parameters, such as frequency, voltage, impedance, phase
angle, active power, reactive power, and harmonic distortion at
the DG site for islanding detection. These techniques are further
categorized into passive, active, and hybrid techniques. The num-
ber of proposed active and passive techniques increases rapidly
over the last few years [9].
2.3. Passive islanding detection techniques
Passive islanding detection techniques basically monitor the
system parameters, such as frequency, voltage, and harmonics at
the point of common coupling, or at the DG terminals, and com-
pare it with a predetermined threshold value for islanding detec-
tion. Fig. 3 shows the basic working principle of passive islanding
detection techniques.
Some of the most common passive techniques are the rate of
change of power (ROCOP) [7], rate of change of frequency
(ROCOF) [19–21], rate of change of frequency over power
(ROCOFOP) [22], change of impedance [23], voltage unbalance
[24], over/under (O/U) voltage and over/under frequency [25], har-
monic distortion (voltage and current) [24,25], phase jump detec-
tion [26], rate of change of voltage and change in power factor [27].
The characteristics of these techniques are summarized in Table 1.
Cost effectiveness, fast detection speed, and no impact on
power quality are some of the major advantages of passive tech-
niques. However, a major problem with this technique is that it
suffers from large non detection zone (NDZ), and it is very difficult
to detect islanding when the generation and load in the islanded
system are closely matched. Furthermore, the setting of threshold
value requires special consideration. Lower threshold settings may
result in nuisance tripping, and if the setting is too high, islanding
may not be detected. Hence, error detection rates are high. These
drawbacks can be overcome by using signal processing techniques
and intelligent classifiers. However, these problems can also be
overcome by active techniques [13,28].
2.3.1. Active islanding detection techniques
Active islanding detection techniques interact with the power
system by introducing perturbations into the system variables,
such as frequency, voltage, currents, and harmonics. Fig 4 shows
the basic working principle of active islanding detection tech-
niques. The impact of these perturbations is significant if the dis-
tributed generation resource is islanded otherwise quite negligible.
Some of the most common active techniques are reactive power
export error detection (RPEED) [29], impedance measurement
[7,23], slip mode frequency shift method (SMS) [30,31], Active fre-
quency drift (AFD) [32], frequency jump (FJ) [13], Active frequency
drift with positive feedback (AFDPF) [33], sandia frequency shift
(SFS) [34,35], sandia voltage shift (SVS) [36], variation of active
and reactive power [11,13,32], negative sequence current injection
[37,38], high frequency signal injection [39,40], virtual capacitor
[41], virtual inductor [42] and phase PLL perturbation method
 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624 615

Islanding Detection

Signal
Local Remote
Processing
Techniques Techniques
Techniques

Passive Active Hybrid

Rate of change of Reactive Power Voltage Unbalance and Transfer Trip

Frequency Set Point
Fourier Transform
Power Export Error Detection Scheme

Rate of change of Impedance Voltage and Real Power

Shift Technique PLCC Scheme Wavelet Transform
Frequency Measurement

Rate of change of Slip Mode Voltage Fluctuation

Frequency over Power Injection
S - Transform
Frequency Shift

Change of Active Frequency Hybrid SFS and Q-f

TT – Transform
Impedance Drift Technique

Hilbert Huang
Voltage unbalance Frequency Jump
Transform

Over/Under Voltage Active Frequency Drift

and Frequency with Positive Feedback

Sandia Frequency
Harmonic Distortion
Shift

Phase Jump Sandia Voltage

Detection Shift

ROCOV and Change Variation of Active

in Power Factor and Reactive Power

Negative Sequence
Current Injection

High Frequency
Signal Injection

Virtual Capacitor

Virtual Inductor

Phase PLL
Perturbation

Fig. 2. Categorization of islanding detection techniques.

[43]. Most of these techniques are used for inverter type DGs. The
characteristics of these techniques are summarized in Table 2.
In comparison to passive techniques, active techniques reduce
the non-detection zone (NDZ) and decrease the error detection
rate. Furthermore, in order to inject perturbations in the power
systems, additional controllers/power electronics equipment is
required. This increases the complexity of the system and reduces
the power quality. Moreover, additional detection time is required
in order to observe the power system response on perturbations.
Hence, the system’s stability is significantly degraded.
616 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624

YES Disconnect DG
Islanding Connected to Local
Load
Measure and
Analyze Is Parameter >
Parameter Signal Threshold Value?
at the PCC Power
Quality
NO Indices

Fig. 3. Basic working principle of passive islanding detection techniques.

Table 1 techniques have the potential to fulfil these requirements, pro-

Characteristics of different passive techniques. vided that their limitation are solved using some other means.
Method Detection Error Impact on NDZ Passive islanding detection techniques have the advantage of not
time detection power quality degrading the power quality, capable of detecting islanding
rate quickly, and are compatible for all types of DGs. However, their
ROCOP 24–26 ms High No Small main limitation is the large non detection zone and threshold set-
ROCOF 24 ms High No Small tings. Both of these limitations can be addressed by using signal
ROCOFOP 100 ms Low No Smaller
processing techniques and intelligent classifiers.
than ROCOF
Change of 10 ms Low No Small
impedance
3. Signal processing techniques for islanding detection
Voltage 53 ms Low No Large
unbalance
O/U voltage and 4 ms to Low No Large Signal processing techniques are commonly used to improve
frequency 2s the performance of passive islanding detection techniques.
Harmonic 45 ms High No Large for Versatility, stability, cost effectiveness, and ease of modification
distortion high Q
properties of the signal processing techniques help researchers
Phase jump 10–20 ms Low No Large
extract the hidden characteristics of the measured signals for
islanding detection. On the basis of these extracted features, deci-
sion can be made on whether islanding did occur. Fig. 6 shows the
2.3.2. Hybrid islanding detection techniques basic steps involved in islanding detection using signal processing
Hybrid islanding detection techniques basically works on the techniques.
combined features of the aforementioned techniques and is effec- The basic signal processing tools, which are used for islanding
tively applied to complex systems. During islanding detection, the detections, are Fourier transform, s-transform, Hilbert Huang
passive technique functions as primary, while the active technique transform, wavelet transform, and tt-transform. The description
functions as a secondary, as shown in Fig 5. Hence, the combina- of these signal processing tools used in islanding detection tech-
tion of these methods will improve the multiple performance niques are discussed in the following sections.
indices.
Some examples of hybrid islanding detection techniques 3.1. Fourier transform based islanding detection techniques
include voltage unbalance and frequency set point method [44],
technique based on voltage and real power shift [45], voltage Fourier transform is the most common technique in frequency
fluctuation injection [46], hybrid SFS and Q–f technique [47], and domain analysis. It basically represents a signal as a summation
the technique based on rate of change of reactive power and load of sinusoidal terms of different frequencies. It extracts the features
connecting strategy [48]. of the stationary signal at specific frequencies, but is incapable of
These techniques possess very small non-detection zone (NDZ), detecting the time distribution of different frequencies. It is also
and degradation in power quality is also reduced because per- unable to resolve any momentary information associated with
turbations are only introduced when islanding is suspected. fluctuations [49]. Hence, the time–frequency analysis is proposed.
However, this combination increases the cost of the system, along The short time Fourier transform (STFT) is the modification of
with the islanding detection time. Fourier transform. It divides the signal into small frames, where
From the above discussion, it is noticed that each islanding each frame can be assumed to be stationary. These numerous
detection technique has its own merits and demerits. However, frames of the signal are evaluated by the moving window. This
the issues that are still unresolved include accuracy, very high moving window identifies the relation between the time and the
detection speed, and compatibility of detecting islanding in multi- change in frequency [50]. However, STFT cannot analyze the non-
ple and hybrid DGs environment. In this regard, it can be noted stationary signal due to the limitation of fixed window’s width
from the aforementioned review that passive islanding detection [51].

YES Disconnect DG
Islanding Connected to
Continuous Measure and Local Load
Injection of Analyze
Is Parameter >
Perturbations at Parameter
Threshold Limit?
specified Interval Signal at the
PCC Power
at the PCC Quality
NO Indices

Fig. 4. Basic working principle of active islanding detection techniques.

 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624 617

Table 2 DFT taken over short time section of the signal. It directly calcu-
Characteristics of different active techniques. lates the amplitude and phase of the desired frequency of the input
Techniques Detection Error Impact on NDZ signal, which ultimately reduces the computational time [53–55].
time detection power Jae-Hung et al. [56] uses Goertzel algorithm to reduce the islanding
rate quality detection time in single phase 2 stage Photovoltaic (PV) system. In
RPEED 2s Low Degrades Small the proposed system, the inverter injects the output current with a
Impedance 0.77– Low Degrades Small ninth harmonic component into the grid, and detects the same in
measurement 0.95 s
SMS 0.4 s Low Degrades Small
voltage at the point of common coupling. NDZ does not exist in this
(approx) method, even under perfect match of power and load. The impact
AFD With 2 s High Degrades Large if on the power quality is also negligible, and islanding is detected
value of Q is within 2 cycles.
high
FJ 75 ms Low Degrades Small
AFDPF 1s Lower Slightly Smaller than 3.2. Wavelet transform based islanding detection techniques
(approx) than AFD degrades AFD
SFS 0.5 s Low Slightly Smallest The wavelet transform (WT) is also used for the analysis of the
degrades
SVS 0.5 s Low Slightly Smallest
signal. It is a mathematical model based on square integral and
degrades group theory, similar to FT. It decomposes a signal into its con-
Variation of active 0.3–0.75 s High Degrades Small stituents at different frequency scales [57,58]. It represents a signal
and reactive in both time and frequency domains. Hence, it is suitable for
power
examining the signals in those applications where time–frequency
Negative sequence 60 ms Low Degrades None
current injection resolution is considered necessary. It has been widely used in
High frequency Few ms Low Slightly Smallest power system applications, such as detection, disturbance transi-
signal injection degrades tion events in electrical power quality, feature extraction, power
Virtual capacitor 20–51 ms Low Slightly Smallest system protection, and de-noising [59–61].
degrades
Virtual inductor 13–59 ms Low Slightly Smallest
The wavelet transform is more advantageous compared to
degrades Fourier based transforms (STFT, FFT and DFT).This is due to the fact
Phase PLL 120 ms Low Negligible Smallest that window size is fixed in Fourier based transform, but varies in
perturbation wavelet transform. Hence, time–frequency resolutions are not
compromised. Furthermore, wavelet transform determines the
time and frequency information simultaneously for low and high
Discrete Fourier transform (DFT) is very dominant technique for frequencies by long and short windows, respectively [62].
frequency domain analysis of discrete time signals. It transforms Wavelet transform is categorised into continuous (CWT) and dis-
the discrete time sequence of finite length into discrete frequency crete wavelet transforms (DWT). CWT is used in islanding detec-
sequence of finite length. Fast Fourier transform (FFT) also gives tion by analysing DG voltage. The Mallat decomposition is also
similar results as the DFT, although in lesser amounts of time. used to extract and eliminate the noise from the signal [63]. This
However, it is not suitable for the analysis of non-stationary sig- method reduces the computational efficiency by introducing the
nals, due to the fact that it depicts those spectral values that do numerous coefficients. This problem diverts the attention of the
not exist in the original signal. Kim [52] introduces a new passive researchers towards DWT.
technique for islanding detection based on DFT for the extraction In [64,65], authors used DWT to analyze the voltage signal from
of the desired features. Conventional passive techniques misinter- a single phase PV system. The proposed method uses Bi-orthogonal
pret the grid disturbances, such as sag, swell, and transient condi- 1.5 and 5 decomposition levels for islanding detection. The reduc-
tions. NDZ might also have occurred if the power generated by tion in the number of sensors, minimisation in computational bur-
photovoltaic (PV) and the load matches. The proposed method uses den and complexity are some of the advantages associated with
the variations in 2nd harmonic components of grid voltage. The this technique. Daubechies mother wavelet based DWT is used in
utilization of harmonic coefficients provides robust control against [66]. It examines the variations in voltage and frequency. The sali-
grid disturbances and also reduces the NDZ. The islanding detec- ent features of this method are the simplicity in programming,
tion time is around 1 ms via the usage of high performance DSP enhancement of islanding detection capability, and simultaneous
controller. observation of power quality profiles. The proposed scheme is
In order to overcome the problem of slow/reduced computation tested and verified in several scenarios with flexibility, feasibility,
time of DFT, Goertzel algorithm is used. It is basically a type of dis- and robustness. In [67], negative sequence of current and voltage
crete Fourier transform and the fastest technique of pitch detection signals are considered by Daubechies db4 based DWT. The stan-
compared to FFT and DFT. It works from the perspective of filtering dard deviation and change in energy coefficients discriminates
operation at a specified frequency, or from the perspective of the between islanding and other disturbance conditions. In the

PASSIVE YES
Disconnect DG
Islanding Connected to
Local Load
Measure and
Analyze Is Parameter >
Parameter Signal Threshold Limit?
at the PCC
Inject ACTIVE
Perturbation
NO
at the PCC

Fig. 5. Basic working principle of hybrid techniques.

618 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624
Parameter Analyze and
Signal (Voltage, Detect Disturbance
current, using Signal
frequency and Processing
harmonics) Techniques/Tools
Fig. 6. Block diagram for signal processing based islanding detection technique.
proposed method, islanding is detected in 1 cycle by using the first
level of energy and standard deviations. Researchers benefit from
the compactness and localisation properties of the Daubechies
db4 in diminishing NDZ. The proposed scheme is compared with
the existing passive (over/under voltage and frequency) scheme,
and found very efficient and effective in all working conditions
[68]. Karegar et al. [69] proposes the DWT for the islanding detec-
tion of wind turbines. Voltage profiles are examined by db5. The
proposed scheme proves dependability under different load condi-
tions. In [70], db4 DWT-based technique is utilised for grid con-
nected PV DGs. The spectral changes in the higher frequency
components of voltage are analyzed for islanding detection. The
proposed scheme is found very effective, and detected islanding
conditions within 0.05 s. Sharma et al. [71] utilises the localisation
and compactness property of the Dyadic wavelet transform for
islanding detection. It has been found that the proposed scheme
discriminates the islanding and non-islanding scenarios for grid
connected PV system in more than 1 cycle. Liu et al. [72] introduces
the islanding detection scheme based on WT for the stand alone
operation of DG system.
The main disadvantages associated with DWT are the integra-
tion of the high frequencies and measurement of several electrical
quantities. Therefore, wavelet packet transform (WPT) is proposed
in [73]. This scheme proposes a new index called node rate of
change of power index. This index computes the change of the
power at each WPT sub band. The base of WPT is db10, because
it has a smaller number of wavelet coefficients. In [74], the
‘‘Haar’’ mother wavelet has been used for islanding detection.
This type of mother wavelet requires the least decomposition
levels, thus, has the least detection time. This method calculates
the current signal at the point of common coupling (PCC) and
detects the islanding within 5.5 ms. The proposed method is also
applicable for multi-DG environment.
In [75], a new feature extraction technique is proposed for
islanding detection. This work examine the variations in harmonic
profiles for inverter based DG system, and is basically an extension
of [65]. Wavelet transform-based multi-resolution analysis (WT-
based MRA) technique is used for feature extraction. The WT-based
MRA decomposes the output voltage into multiple scales. Each
scale produces a sequence of wavelet coefficients (WCs) on the
bases of frequency bandwidth. The change in the ratio of the
WCs is used for islanding detection. The simulation results prove
that the proposed scheme is successful for islanding detection
under all operating conditions.
Samui and Samantaray in [76,77] proposes a new technique for
islanding detection based on wavelet singular entropy (WSE). WSE
integrates the advantages of wavelet transform, singular value
decomposition, and Shanon entropy. In the proposed scheme, the
wavelet transform analyzes the three phase voltage signal and gen-
erates the detailed coefficients. Singular value matrix is calculated
from these detailed coefficients in order to determine the WSE for
each phase. Finally, WSE index is produced by adding all the phases
of WSE. The proposed scheme is compared with the two existing
schemes (rate of change of frequency (ROCOF) and rate of change
of power (ROCOP)) in a perfectly matching environment, and it
YES
Islanding
Fix
Is Parameter >
Threshold Threshold Value?
Value
Other
Disturbance
NO
has been found that the proposed scheme detects islanding condi-
tion effectively within 10 ms.
3.3. s-transform based islanding detection techniques
Wavelet transform extracts the desired features of the signal
from both time and frequency domains. However, batch processing
and noise sensitivity are the disadvantages associated with this
technique. To overcome these problems, Stockwell proposes s-
transform in 1996. In this technique, the properties of both STFT
and WT are merged. It is a time–frequency technique having a vari-
able window of STFT and an expansion of WT. It is based on a scal-
able localising Gaussian window, and supplies the frequency
dependent resolution [78–80]. It provides multi-resolution, and
keeps the phase of each frequency component unchanged. It trans-
forms the signal from the time domain to two dimensional fre-
quency domains. Local spectral characteristics are examined
either by the amplitude time frequency spectrum, or by the phase
time frequency spectrum. In this tool, the sinusoid is fixed with
respect to the time axis, and localised scalable Gaussian window
examined it more comprehensively by dilating and translating
the phase frequency [81,82]. s-transform (ST) provides con-
siderable and noteworthy results in the detection and localisation
of disturbances due to islanding or any other condition. Thus, it
detects the disturbance signal efficiently and accurately.
In [83], the islanding detection technique using s-transform is
proposed for the hybrid system. It clearly shows the domination
of s-transform in comparison to wavelet transform for islanding
detection on the bases of the simulation results. Ray et al. [84,85]
extracts the negative sequence voltage for islanding detection.
They also compared the wavelet transform and s-transform under
noisy conditions. On the basis of the simulation results, it has been
found that s-transform determine the islanding scenario profi-
ciently, within 26–28 ms. Islanding detection, on the basis of per-
formance indices, was also conducted, which is another
contribution to this field. A new technique that relies on s-trans-
form based cumulative sum detector (CUSUM) is proposed in
[86]. In this proposed technique, the spectral energy contents of
the negative sequence voltage and current signals are used for
computation. It has been verified that the proposed technique
detects the islanding condition in 25 ms with an accuracy of more
than 92%.
The problem associated with s-transform is that its perfor-
mance weakens under certain operating situations, such as tran-
sients. The s-transform is modified to hyperbolic s-transform to
overcome the adverse effect of transients and to realize a superior
signal processing technique [87,88]. Compared to s-transform, the
hyperbolic s-transform has a pseudo Gaussian hyperbolic window.
It has frequency dependence in its shape in integration to its width
and height. This asymmetrical window provides better resolution
in both time and frequency at high and low frequencies, respec-
tively. In [89], hyperbolic s-transform is used to detect the island-
ing condition. The change in energy and standard deviation (STD)
of the perturbed voltage signal at PCC is resolute, and on the basis
of these predicated values; a suitable threshold is finalised in order
 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624
to detect an islanding event. The results clearly reveal the advan-
tages of the proposed scheme for islanding detection under both
noise and noise-free environments.
3.4. tt-transform based islanding detection techniques
Many time varying frequency techniques are used to process
non-stationary signals. Some of the most common techniques are
STFT, wavelet, and s-transform. However, these transforms intro-
duce redundancy, passing from a 1-D (one-dimensional) time sig-
nal to a 2-D time–frequency (or time-scale) signal. In 2003, an
incipient technique based on s-transform is proposed, called the
time-time transform (tt-transform). It includes redundancy in time
passing from a 1-D time signal to a 2-D time-time signal [90,91].
Aziah et al. [92] proposed a new signal processing tool i.e. tt-
transform for islanding detection. The results of the proposed
scheme were represented; utilizing 2-D TT transformed plots of
the original signals. It was confirmed that the technique is capable
of detecting the islanding scenario in a more precise and expedi-
tious manner, because each event possess a distinctive/unique pat-
terns. In [89], tt-transform is used to extract the desired features
for islanding detection. The results obtained are compared with
wavelet and s-transform. It is found that the proposed technique
is superior for islanding detection in all conditions.
3.5. Hilbert Huang transform based islanding detection techniques
The Hilbert Huang transform (HHT) is a novel signal processing
technique. It consists of two different procedures. At the first stage,
the signal to be examined is decomposed into intrinsic mode func-
tion (IMF) that have consequential instantaneous frequencies and
amplitudes, via utilization of the empirical mode decomposition
(EMD) process. At the second stage, the IMFs are sorted from the
highest frequency to the lowest frequency. The Hilbert transform
can then be applied to each IMF, giving the instantaneous ampli-
tude and instantaneous frequency versus time curve. This
combination of EMD process and Hilbert transform is known as
the HHT [93,94]. The dominance of this scheme over wavelet trans-
form, STFT, and s-transform has been presented in literature [95–
97].
In [98], an incipient passive islanding detection scheme for
inverter-based DGs is offered, which utilises HHT for features
extraction. The simulation results showed that the proposed
scheme can efficaciously detect islanding in less than two cycles.
Moreover, the simplicity, efficacy, expeditiousness, and robustness
of the technique against noise are verified in multi-DG systems.
Table 3 summarises the aforementioned signal processing based
islanding detection techniques.
4. Signal processing based islanding detection techniques with
intelligent classifier
Until now, the islanding detection techniques relying solely on
the signal processing tools are discussed. In signal processing
based islanding detection techniques, the desired features are
extracted from the input signal and compared to a threshold value.
The selection of threshold value is quite a difficult task. If its value
is set high, then islanding will not be detected, while if it is set very
low, then it trips the DG even for the case of disturbances. To over-
come this issue, intelligent classifiers have been combined with
signal processing based islanding detection techniques.
Intelligent classifiers commonly used in signal processing based
islanding detection techniques are decision tree (DT), artificial neu-
ral network (ANN), probabilistic neural network (PNN), adaptive
neuro fuzzy inference system (ANFIS), random forest (RF), support
619
vector machine (SVM) and Fuzzy logic control. These intelligent
classifiers enhanced the efficiency, speed, accuracy, and can detect
islanding without using any threshold settings, as in the case of
common signal processing-based passive techniques. The basic
theme of this scheme is shown in Fig. 7.
Guiliang [99] presented a new technique for islanding detection
based on FFT for feature extraction and artificial immune system
(AIS) as intelligent classifier, respectively. On the basis of the sim-
ulation results, it was verified that the proposed scheme was very
efficient, and requires very advanced digital signal processor (DSP)
for it to be implemented. In [100] and [101], the researchers used
an intelligent classifier, along with the DFT to check the efficiency
and reliability of the system. Kar and Samantaray [100], derives 27
features through DFT pre-processor in order to train the data min-
ing model. The data mining model consist of decision tree (DT),
random forest (RF), and support vector machine (SVM). The pro-
posed scheme is tested by taking into account the inverter and syn-
chronous based DG in the micro-grid. This paper has compared the
accuracy of these intelligent classifiers. The accuracy of SVM and
RF is very close to DT, but the implementation of DT on DSP/
FPGA is quite easy compared to SVM and RF. The proposed scheme
detects the islanding condition in less than 1.5 cycles. This DT
model is also compared with [102], which took into account 11
features for islanding detection. It is concluded that the proposed
scheme is more comprehensive in making assessments, and ser-
iously impact decision boundaries. The proposed scheme also pro-
vides a more generalised solution for both synchronous and
inverter based DG compared to the existing intelligent anti-island-
ing models based only on synchronous based DG [103]. Abd-
Elkader et al. [101] processes the voltage and current signals with
DFT to extract the 2nd harmonic components. These components
are fed to artificial neural network (ANN) for decision making.
This proposed passive scheme for double fed induction generator
(DFIG) wind turbines detect the islanding condition within 2
cycles, and has no NDZ if the load values are within the prescribed
limits.
Lidula et al. [104] proposed a novel pattern recognition
approach for fast islanding detection. DWT is used to extract the
desired features of transient voltage and current signals, and to
train the decision tree (DT) classifier for islanding detection. The
proposed scheme is tested on a medium voltage distribution sys-
tem with multiple DGs, and it detects islanding within 24 ms.
The same scheme is further tested in [105] for synchronous and
induction type DGs. In this case, it detects islanding condition
within two cycles, with more than 98% accuracy. The proposed
method is again checked for VSC-based DG and induction genera-
tor. In this scenario, the islanding is detected within 3 cycles.
Furthermore, it is also robust, despite the fact that the voltage
and current profiles are riddled with noise signals [106].
In [107], a novel technique is proposed based on ANFIS classifier
and ‘Haar’ mother wavelet based DWT. The proposed technique
reduces the NDZ to zero along with the islanding detection within
prescribe limits. Heidari et al. [108] uses DWT and decision tree
(DT) to examine the voltage transient signal for islanding detec-
tion. It is verified that the proposed scheme detects the islanding
condition within one cycle. Simplicity, speed, low cost and high
accuracy are the merits associated with this technique. In [109],
DWT and artificial neural network (ANN) based islanding detection
technique is proposed. It is indicated that the proposed method
detects the islanding condition with high accuracy. Furthermore,
it is also proven that the proposed technique is superior to the
commonly used under and over voltage and frequency (UFP/OFP
and UVP/OVP) techniques.
Although s-transform (ST) has the potential to assess the power
signals perturbances, it requires more computational time to pro-
cess the signal. Thus, the conventional ST is unsuitable for
620 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624

Table 3
Summary of different signal processing based islanding detection techniques.

Category Reference SP method Tested DG system Detection Advantages & disadvantage

time
Fourier Kim [52] DFT PV Around 1 ms Robust control against grid disturbance
transform Jae-Hyung [56] DFT (Goertzel Single phase 2 stage PV PCS Less than 2 Fast harmonic computation
Algorithm) cycles
Wavelet Yanping [63] CWT Grid connected DG 0.6 s Low computational efficiency
transform Pigazo [64,65] DWT Low voltage and low power PV Vary from 0.5 Less complex. Minimum computational burden
system to 30 cycles
Hsieh [66] DWT DG installed on petroleum company 0.05–0.1 s Highly effective
Samantaray [67] DWT Wind farm (DFIG) 1 cycle Highly efficient
Hanif [68] DWT Grid connected PV system 2.5 frequency Highly efficient and diminishes NDZ
cycles
Karegar [69] DWT IG type wind turbine Less than 0.2 s More reliable
Hanif [70] DWT PV DG system 0.05 s Effective, less burden on processor and very fast. But it
not worked in harmonic free environment
Sharma [71] DWT Grid connected PV system More than 1 Efficient and effective
cycle
Morsi [73] Wavelet Wind farm (9 MW) 0.06 s Simplicity, computationally efficient and easy
Packet implementation with high accuracy
transform
Shariatinasab [74] DWT Synchronous DGs 5.5 ms Works efficiently with high accuracy, compatible to
multi DG environment and NDZ did not occur
Jiaxin [75] WT-based Inverter based DG –
MRA
Samui [76,77] WSE 3 wind farms and 1 emergency 10 ms Effective, easily implemented on DSP but NDZ occurs if
diesel generator power mismatch is above 15%
s-transform Ray [84,85] s-transform Inverter based & rotating machine 26–28 ms s-transform is most efficient but WT fails under noisy
based i.e. PV, Fuel cell and wind condition
Samantaray [86] s-transform Wind Farm (DFIG) 25 ms Performance accuracy is more than 92%. Highly
based CUSUM effective for large power system with multiple DGs
Mohanty [89] Hyperbolic s- Hybrid (PV, Fuel cell, wind) – Best under noisy environment
transform
tt-transform Mohanty [89] tt-transform Hybrid (PV, Fuel cell, wind) – Best under noisy environment
Aziah [92] tt-transform 1200Vdc DG system – Fast and accurate
Hilbert Mohammadzadeh HHT Inverter based DG Less than 2 Efficient, simple and robust against noise
Huang [98] (EMD + IMF) cycles
transform

Islanding
Parameter Detect Disturbance
Desired Decision by
Signal (Voltage, using Signal
Feature Intelligent
current, Processing Classifier
Extraction
frequency) Techniques
Other
Disturbance

Fig. 7. Basic block diagram of SP based technique with intelligent classifier.

authentic-time applications, unless its speed is significantly incre-
mented. There have been some methods that reduces the
computational time for the calculation of discrete ST, such as
Generalized Fourier family transform (GFT) [110]. The GFT algo-
rithm amalgamates down sampling and signal cropping to gener-
ate a discrete Fast S-transform (DFST). Such a scheme diminishes
the retrieval of unwanted information, thereby constraining the
computational requisites. Dash et al. [111] proposed an incipient
islanding detection method predicated on DFST and Fuzzy system.
Both the negative sequence voltage and currents are quantified at
the DG location, which are utilised as inputs to the DFST processing
module, resulting in features like spectral energy and standard
deviation. For detecting power islands, the features from DFST
exhibiting consequential fluctuations are given as inputs to the
Fuzzy classifier for the relegation of a non-islanding and islanding
event. Upon utilising the proposed scheme on different dis-
tribution network, it is verified that the detection time is less than
a cycle.
A comparative study on some of the signal processing tech-
niques for islanding detection has been presented in [112]. The
proposed scheme extracts the negative sequence components of
voltage signal using hyperbolic s-transform (HST), tt-transform
(TTT), and mathematical morphology methods. The decision
between islanding and non-islanding conditions is made by sup-
port vector machine (SVM). It is presented that HST, TTT, and
mathematical morphology methods are more accurate compared
to commonly used ST and WT. Furthermore, the proposed tech-
nique also works effectively in noisy and noise-free environments.
Table 4 summarises the signal processing based islanding detec-
tion techniques with intelligent classifiers.
5. Miscellaneous feature extraction techniques
Apart from these techniques, there are some other signal pro-
cessing techniques playing a paramount role in feature extraction
for islanding detection. In [113,114], the phase space technique is
used to extract the desired features from the signal. In [114], radial
basis function (RBF) and probabilistic neural network (PNN) classi-
fiers are utilised for decision-making in combination with phase
space technique. On the basis of the results, it is verified that
 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624 621

Table 4
Summary of different SP techniques with intelligent classifier.

Category Reference SP Intelligent Tested DG system Detection Advantages & disadvantage

method classifier time
Fourier Guiliang FFT AIS Inverter based DG – Very efficient but implementation requires advanced DSP
transform [99] because of high computational complexity
Kar [100] DFT Data mining Synchronous and Inverter Less than Very fast
model (DT, RF, based DGs 1.5 cycles
SVM)
Abd- DFT ANN Wind farm DFIG Within 2 NDZ will occurs if load value is far from prescribed limits
Elkader cycles
[101]
Wavelet Shayeghi DWT ANFIS Voltage source converter based 1 s approx Overcome the problem of setting the threshold
transform [107] DG’s
Arachchige DWT Decision tree Multiple DG’s 24 ms Very fast
[104]
Lidula DWT Decision tree Induction and synchronous Below 2 Accuracy is more than 98%
[105] generators cycles
Lidula DWT Decision tree VSC based DC source and Below 3 High dependability, higher security, faster response time,
[106] induction generator cycles less sensitivity to load imbalance and robust against noise
Heidari DWT Decision tree Induction and synchronous Lower Simplicity, speed, less costly and higher Accuracy (98%)
[108] generators than 1
cycle
Fayyad DWT ANN PV Panel connected to three – High accuracy, show excellent performance for high Q
[109] phase inverter load.
s-transform Dash [111] DFST Fuzzy rule based Synchronous generator, DFIG Less than High accuracy, reliability, detection rate and sensitivity
classifier wind turbines, PV system and a cycle
CHP
Phase space Khamis Phase PNN 1200Vdc DG’s 0.2 s 100% accuracy
technique [113] space

Table 5
Comparison of features among different islanding detection techniques.

Characteristics Remote techniques Local techniques Signal processing based

techniques
Passive Active Hybrid
Principle Communication between utility Monitor natural effects of Inject Combination of both passive Monitor changes in system
grid and DGR islanding disturbances and active parameters
Detection time Short Short Long Long Very short
Impact on power No No High Very small No
quality
Non detection zone None Large Small Very small No
Reliability High Low High High Very high
Cost High Low Low Low Low
Effect on distribution None None Direct Lower than active None
system influence
Multiple DGs Possible Possible Not possible Possible Possible
operation

PNN is better and works more efficiently than RBF. In [113], perfor-
mance evaluation has been conducted between the radial basis
function neural network (RBFNN) and probabilistic neural network
(PNN) using phase space technique as a signal extractor. The
results proved that the PNN classifier is more superior to RBFNN,
and has an accuracy of 100%.
Zeineldin et al. [6] presented a new islanding detection tech-
nique for synchronous DGs based on total least square estimation
of signal parameters via rotational invariance techniques (TLS-
ESPRIT). This new technique is predicated on two incipient
parameters for islanding detection; the frequency of oscillation
and the damping factor of the DG frequency output. Negligible
NDZ is a paramount advantage of this technique.
6. Discussion
The main focus of this paper is to discuss the capability of signal
processing techniques used for accurate and precise detection of
islanding condition within the shortest period of time. On the basis
of critical review, it has been found that the signal processing
based techniques are the preferable option for islanding detection
compared to the passive, active, and hybrid techniques. The use of
features with higher discriminating power is the main ingredient
that boosts the performance of signal processing techniques with
intelligent classifiers. In detecting islanding condition, their perfor-
mance surpasses those of the conventional techniques in terms of
detection time and accuracy without affecting power quality.
Furthermore, they can also overcome the non-detection zone and
threshold setting that have been the Achilles heels of conventional
techniques. These techniques also have the potential to work in
multiple DGR’s environment with very high reliability. A perfor-
mance comparison among remote, local, and signal processing
based techniques are summarized in Table 5.
From Table 5, it can be observed that signal processing based
techniques are more efficient and reliable compared to the existing
islanding detection techniques. Although signal processing based
techniques use similar parameters as the passive techniques, they
622 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624

Table 6 Acknowledgement
Comparison between different feature extraction techniques.

Sr. Feature Merits Demerits This work is supported by the Ministry of Education Malaysia
No extraction (HIR-MOHE D000004-16001), University of Malaya and ‘‘SATU
technique
Joint Research Scheme 2014’’.
1 Discrete Widely used to analyze the Computational time is
Fourier non-stationary signal slow
References
transform
(DFT)
[1] Tsang WLCKM. Three-level grid-connected photovoltaic inverter with
2 Wavelet Provide local representation Batch processing,
maximum power point tracking. Energy Convers Manage 2013;65:221–7.
transform in both time and frequency integration of high
[2] Gsänger S. World Wind Outlook: Down But Not Out; 2011. <http://www.
(Variable window size). High frequencies and greatly renewableenergyworld.com/rea/news/article/2011/05/world-wind-outlook-
computational speed influenced by noise down-but-not-out>.
3 s-transform Convertibility of signal from Its performance [3] Soma Biswas SKG, Chatterjee Amitava. Optimum distributed generation
time domain to 2- weakens in the presence placement with voltage sag effect minimization. Energy Convers Manage
dimensional frequency of transients and 2012;53:163–74.
domain. Efficient and harmonics. Does not [4] Banaei ESMR. Application of multi-stage converter in distributed generation
accurate work properly in real systems. Energy Convers Manage 2012;62:76–83.
time environment [5] IEEE Std 929-2000 (Revision of IEEE Std 929–1988). IEEE Recommended
4 tt- It includes redundancy in Resolution at the time of Practice for Utility Interface of Photovoltaic (PV) Systems, USA, IEEE Std 929-
transform time passing from 1-D time initiation of event 2000.
signal to 2-D time signal. requires improvement [6] Zeineldin HH, Abdel-Galil T, El-Saadany EF, Salama MMA. Islanding detection
of grid connected distributed generators using TLS-ESPRIT. Electr Pow Syst
Amplifies high frequency
Res 2007;77:155–62.
components against low
[7] Ku Ahmad KNE, Selvaraj J, Rahim NA. A review of the islanding detection
frequency components
methods in grid-connected PV inverters. Renew Sust Energy Rev
5 Hilbert Novel feature extraction Fail to produce accurate 2013;21:756–66.
Huang technique for distorted frequency spectrum, loss [8] Mahat P, Zhe C, Bak-Jensen B. Review on islanding operation of distribution
transform signal. Robust and rigorous of frequency system with distributed generation. IEEE Power Energy Soc Gen Meeting
components that have 2011:1–8.
low energy contents [9] Khamis A, Shareef H, Bizkevelci E, Khatib T. A review of islanding detection
techniques for renewable distributed generation systems. Renew Sust Energy
Rev 2013;28:483–93.
[10] Wei CWT, Teoh Yee. An overview of islanding detection methods in
photovoltaic systems. World Acad Sci, Eng Technol 2011;5(10):533–41.
[11] Velasco D, Trujillo CL, Garcerá G, Figueres E. Review of anti-islanding
achieve higher precision in dealing with complex systems, due to
techniques in distributed generators. Renew Sust Energy Rev
their versatility, stability, and robustness. 2010;14:1608–14.
The use of highly discriminating features extracted by Fourier [12] Yu B, Matsui M, Yu G. A review of current anti-islanding methods for
transform, wavelet transform, s-transform, tt-transform, and photovoltaic power system. Sol Energy 2010;84:745–54.
[13] Li C, Cao C, Cao Y, Kuang Y, Zeng L, Fang B. A review of islanding detection
Hilbert Huang transform have enabled signal processing based methods for microgrid. Renew Sust Energy Rev 2014;35:211–20.
islanding detection techniques to distinguish islanding condition [14] Balaguer-Álvarez IO-R. E.I. survey of distributed generation islanding
and other power quality phenomena more accurately. The benefits detection methods. Revista IEEE Am Latina Trans 2010;8:565–70.
[15] Yin Jun, Chang Liuchen, Diduch C. Recent developments in islanding
and drawbacks of several feature extraction techniques used for detection for distributed power generation. In: Large engineering systems
islanding detection are summarised in Table 6. conference on power engineering (LESCOPE-04); 2004. p. 124–8.
From an overall perspective, it can be seen that hyperbolic s- [16] Wilsun X, Guibin Z, Chun L, Wencong W, Guangzhu W, Kliber J. A power line
signaling based technique for anti-islanding protection of distributed
transform and tt-transform are more efficient and accurate com- generators; Part I: Scheme and analysis. IEEE Trans Power Deliver
pared to other feature extraction techniques [89,112]. 2007;22:1758–66.
[17] 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
7. Conclusion 2007;22:1767–72.
[18] Barsali S, Ceraolo M, Pelacchi P, Poli D. Control techniques of dispersed
generators to improve the continuity of electricity supply. IEEE Power Eng Soc
This paper presents a comprehensive review on signal process- 2002:789–94.
ing techniques used in islanding detection. First, an overview of the [19] Ding X, Crossley PA. Islanding detection for distributed generation. IEEE
conventional techniques outlining their advantages and weak- Power Tech, Russia 2005:1–4.
[20] Freitas W, Wilsun X, Affonso CM, Zhenyu H. Comparative analysis between
nesses are given. This is followed by a description of well-known ROCOF and vector surge relays for distributed generation applications. IEEE
signal processing techniques that have been incorporated into Trans Power Deliver 2005;20:1315–24.
the conventional methods to improve their performance and over- [21] 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.
come their inherent limitations. The application of intelligent clas- [22] Pai FS, Huang SJ. A detection algorithm for islanding-prevention of dispersed
sifiers, along with signal processing techniques to achieve even consumer-owned storage and generating units. IEEE Trans Energy Convers
higher accuracy, faster detection, and better compatibility are also 2001;16:346–51.
[23] O’Kane P, Fox B. Loss of mains detection for embedded generation by system
discussed. It is worth mentioning that the application of signal pro- impedance monitoring. In: Sixth international conference on developments
cessing tools and intelligent classifiers entails no negative side in power system protection (Conf Publ No 434); 1997. p. 95–8.
effects on the power quality and system reliability. For passive [24] Sung-Il J, Kwang-Ho K. An islanding detection method for distributed
generations using voltage unbalance and total harmonic distortion of
islanding detection techniques, the incorporation of signal process- current. IEEE Trans Power Deliver 2004;19:745–52.
ing tools and intelligent classifiers has helped overcome their lim- [25] De Mango F, Liserre M, Aquila AD, Pigazo A. Overview of anti-islanding
itations of large non detection zone and threshold setting. Efficient algorithms for PV systems. Part I: Passive Methods. In: 12th power electronics
and motion control conference (EPE-PEMC); 2006. p. 1878–83.
signal processing based islanding detection techniques play an
[26] Singam B, Hui LY. Assessing SMS and PJD schemes of anti-islanding with
important role for a successful islanding operation of DGRs. Their varying quality factor. In: IEEE international power and energy conference
implementation can enhance the reliability and power quality of (PECon); 2006. p. 196–201.
the system, and is thus preferable over the conventional islanding [27] 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
detection techniques. Thus, signal processing based techniques factors. In: 7th international conference on developments in power system
have great potential to be utilised in real time applications. protection; 2001. p. 82–5.
 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624
[28] Laghari JA, Mokhlis H, Karimi M, Bakar AHA, Mohamad H. Computational
intelligence based techniques for islanding detection of distributed
generation in distribution network: a review. Energy Convers Manage
2014;88:139–52.
[29] Chowdhury SP, Chowdhury S, Crossley PA. Islanding protection of active
distribution networks with renewable distributed generators: a
comprehensive survey. Electr Power Syst Res 2009;79:984–92.
[30] Lopes LAC, Huili S. Performance assessment of active frequency drifting
islanding detection methods. IEEE Trans Energy Convers 2006;21:171–80.
[31] Liu F, Kang Y, Zhang Y, Duan S, Lin X. Improved SMS islanding detection
method for grid-connected converters. IET Renew Power Gen 2010;4:36–42.
[32] De Mango F, Liserre M, Aquila AD. Overview of anti-islanding algorithms for
PV systems. Part II: active methods. In: 12th international power electronics
and motion control conference (EPE-PEMC); 2006. p. 1884–9.
[33] 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.
[34] Zeineldin HH, Kennedy S. Sandia frequency-shift parameter selection to
eliminate nondetection zones. IEEE Trans Power Deliver 2009;24:486–7.
[35] Zeineldin HH, Conti S. Sandia frequency shift parameter selection for multi-
inverter systems to eliminate non-detection zone. IET Renew Power Gen
2011;5:175–83.
[36] Trujillo CL, Velasco D, Figueres E, Garcerá G. Analysis of active islanding
detection methods for grid-connected microinverters for renewable energy
processing. Appl Energy 2010;87:3591–605.
[37] 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.
[38] Tuyen ND, Fujita G. Negative-sequence current injection of dispersed
generation for islanding detection and unbalanced fault ride-through. In:
46th international universities’ power engineering conference (UPEC); 2011.
p. 1–6.
[39] 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.
[40] 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.
[41] 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.
[42] 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 Gen 2007;1:175–81.
[43] 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.
[44] Menon V, Nehrir MH. A hybrid islanding detection technique using voltage
unbalance and frequency set point. IEEE Trans Power Syst 2007;22:442–8.
[45] 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 Deliver
2009;24:764–71.
[46] Wen-Yeau C. A hybrid islanding detection method for distributed
synchronous generators. In: International power electronics conference
(IPEC); 2010. p. 1326–30.
[47] Vahedi H, Noroozian R, Jalilvand A, Gharehpetian GB. Hybrid SFS and Q–f
islanding detection method for inverter-based DG. In: IEEE international
conference on power and energy (PECon); 2010. p. 672–6.
[48] 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.
[49] Karimi M, Mokhtari H, Iravani MR. Wavelet based on-line disturbance
detection for power quality applications. IEEE Trans Power Deliver
2000;15:1212–20.
[50] Dash PK, Panigrahi KB, Sahoo DK, Panda G. Power quality disturbance data
compression, detection, and classification using integrated spline wavelet
and s-transform. IEEE Trans Power Deliver 2003;18:595–600.
[51] Gu YH, Bollen MHJ. Time-frequency and time-scale domain analysis of
voltage disturbances. IEEE Trans Power Deliver 2000;15:1279–84.
[52] Kim I-S. Islanding detection technique using grid-harmonic parameters in the
photovoltaic system. Energy Procedia 2012;14:137–41.
[53] Gonzalez SA, Garcia-Retegui R, Benedetti M. Harmonic computation
technique suitable for active power filters. IEEE Trans Ind Electron
2007;54:2791–6.
[54] Jacobsen E, Lyons R. The sliding DFT. IEEE Signal Proc Mag 2003;20:74–80.
[55] Sozanski KP. Sliding DFT control algorithm for three-phase active power
filter. In: 21st annual IEEE applied power electronics conference and
exposition (APEC); 2006. p. 7.
[56] Jae-Hyung K, Jun-Gu K, Young-Hyok J, Yong-Chae J, Chung-Yuen W. An
islanding detection method for a grid-connected system based on the
goertzel algorithm. IEEE Trans Power Electr 2011;26:1049–55.
[57] Graps A. An introduction to wavelets. IEEE Comput Sci Eng 1995;2:50–61.
[58] Chen S. Feature selection for identification and classification of power quality
disturbances. In: IEEE Power Eng Soc Gen Meeting, vol. 3; 2005. p. 2301–6.
[59] Daubechies I. The wavelet transform, time-frequency localization and signal
analysis. IEEE Trans Inform Theory 1990;36:961–1005.
623
[60] Santoso S, Powers EJ, Grady WM, Hofmann P. Power quality assessment via
wavelet transform analysis. IEEE Trans Power Deliver 1996;11:924–30.
[61] Santoso S, Powers EJ, Grady WM. Power quality disturbance data
compression using wavelet transform methods. IEEE Trans Power Deliver
1997;12:1250–7.
[62] Morsi WG, El-Hawary ME. Novel power quality indices based on wavelet
packet transform for non-stationary sinusoidal and non-sinusoidal
disturbances. Electr Power Syst Res 2010;80:753–9.
[63] Yanping Z, Qiuxia Y, Junjuan W, Dezhong Z, Yuexin T. A novel islanding
detection method of distributed generator based on wavelet transform. In:
International conference on electrical machines and systems (ICEMS); 2008.
p. 2686–8.
[64] Pigazo A, Moreno VM, Liserre M, Aquila AD. Wavelet-based islanding
detection algorithm for single-phase photovoltaic (PV) distributed
generation systems. In: IEEE international symposium on industrial
electronics (ISIE); 2007. p. 2409–13.
[65] Pigazo A, Liserre M, Mastromauro RA, Moreno VM, Dell’Aquila A. Wavelet-
based islanding detection in grid-connected PV systems. IEEE Trans Ind Electr
2009;56:4445–55.
[66] 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.
[67] Samantaray SR, Pujhari TM, Subudhi BD. A new approach to islanding
detection in distributed generations. In: International conference on power
systems (ICPS); 2009. p. 1–6.
[68] Hanif M, Dwivedi UD, Basu M, Gaughan K. Wavelet based islanding detection
of DC–AC inverter interfaced DG systems. In: 45th international universities
power engineering conference (UPEC); 2010. p. 1–5.
[69] Karegar HK, Sobhani B. Wavelet transform method for islanding detection of
wind turbines. Renew Energy 2012;38:94–106.
[70] Hanif M, Basu M, Gaughan K. Development of EN50438 compliant wavelet-
based islanding detection technique for three-phase static distributed
generation systems. IET Renew Power Gen 2012;6:289–301.
[71] Sharma R, Singh P. Islanding detection and control in grid based system using
wavelet transform. In: 5th IEEE power india conference; 2012. p. 1–4.
[72] Liu YH, Luor TS, Huang SJ, Lin JM. Method and system for detecting stand-
alone operation of a distributed generating system (US20050146224 A1).
Patents; 2005.
[73] Morsi WG, Diduch CP, Chang L. A new islanding detection approach using
wavelet packet transform for wind-based distributed generation. In: 2nd IEEE
international symposium on power electronics for distributed generation
systems (PEDG); 2010. p. 495–500.
[74] Shariatinasab R, Akbari M. New islanding detection technique for DG using
discrete wavelet transform. In: IEEE International conference on power and
energy (PECon); 2010. p. 294–9.
[75] Jiaxin N, Caisheng W. Feature extraction for islanding detection using wavelet
transform-based multi-resolution analysis. In: IEEE Power & Energy Soc Gen
Meeting; 2012. p. 1–6.
[76] Samui A, Samantaray SR. Performance assessment of wavelet transform
based islanding detection relay. In: Annual IEEE India conference (INDICON);
2012. p. 545–50.
[77] Samui A, Samantaray SR. Wavelet singular entropy-based islanding detection
in distributed generation. IEEE Trans Power Deliver 2013;28:411–8.
[78] Ventosa S, Simon C, Schimmel M, Danobeitia JJ, Manuel A. The ‘‘S’’-Transform
from a wavelet point of view. IEEE Trans Signal Process 2008;56:
2771–80.
[79] Stockwell RG, Mansinha L, Lowe RP. Localization of the complex spectrum:
the S transform. IEEE Trans Signal Process 1996;44:998–1001.
[80] Dash PK, Panigrahi KB, Panda G. Power quality analysis using s-transform.
IEEE Trans Power Deliver 2003;18:406–11.
[81] Dehghani MJ. Comparison of s-transform and wavelet transform in power
quality analysis. World Acad Sci, Eng Technol 2009;3:395–8.
[82] Mishra S, Bhende CN, Panigrahi KB. Detection and classification of power
quality disturbances using s-transform and probabilistic neural network. IEEE
Trans Power Deliver 2008;23:280–7.
[83] Ray PK, Mohanty SR, Kishor N, Dubeya HC. Coherency determination in grid-
connected distributed generation based hybrid system under islanding
scenarios. In: IEEE international conference on power and energy (PECon);
2010. p. 85–8.
[84] Ray PK, Kishor N, Mohanty SR. s-transform based islanding detection in grid-
connected distributed generation based power system. In: IEEE international
energy conference and exhibition (EnergyCon); 2010. p. 612–7.
[85] Ray PK, Mohanty SR, Kishor N. Disturbance detection in grid-connected
distributed generation system using wavelet and s-transform. Electr Power
Syst Res 2011;81:805–19.
[86] Samantaray SR, Samui A, Babu BC. s-transform based cumulative sum
detector (CUSUM) for islanding detection in distributed generations. In:
Joint international conference on power electronics, drives and energy
systems (PEDES), India; 2010. p. 1–6.
[87] Biswal B, Dash PK, Panigrahi BK. Non-stationary power signal processing for
pattern recognition using HS-transform. Appl Soft Comput 2009;9:107–17.
[88] Nantian H, Dianguo X, Xiaosheng L. Power quality disturbances recognition
based on HS-transform. In: 1st International conference on pervasive
computing signal processing and applications (PCSPA); 2010. p. 311–4.
[89] Mohanty SR, Kishor N, Ray PK, Catalao JPS. Islanding detection in a distributed
generation based hybrid system using intelligent pattern recognition
624 S. Raza et al. / Energy Conversion and Management 96 (2015) 613–624
techniques. In: 3rd IEEE PES international conference and exhibition on
innovative smart grid technologies (ISGT), Europe; 2012. p. 1–5.
[90] Pinnegar CR, Mansinha L. A method of time–time analysis: The tt-transform.
Digit Signal Process 2003;13:588–603.
[91] Simon C, Schimmel M, Danobeitia JJ. On the tt-transform and its diagonal
elements. IEEE Trans Signal Process 2008;56:5709–13.
[92] Aziah Khamis HS, Wanik MZC. Pattern recognition of islanding detection
using tt-transform. J Asian Sci Res 2010;2:607–13.
[93] Afroni MJ, Sutanto D, Stirling D. Analysis of nonstationary power-quality
waveforms using iterative Hilbert Huang transform and SAX algorithm. IEEE
Trans Power Deliver 2013;28:2134–44.
[94] Drummond CF, Sutanto D. Classification of power quality disturbances using
the iterative Hilbert Huang transform. In: 14th international conference on
harmonics and quality of power (ICHQP); 2010. p. 1–7.
[95] Peng ZK, Tse PW, Chu FL. A comparison study of improved Hilbert-Huang
transform and wavelet transform: application to fault diagnosis for rolling
bearing. Mech Syst Signal Process 2005;19:974–88.
[96] Donnelly D. The fast Fourier and Hilbert-Huang transforms: a comparison. In:
Multiconference on computational engineering in systems applications
(IMACS); 2006. p. 84–8.
[97] Ayenu-Prah AY, Attoh-Okine NO. Comparative study of Hilbert-Huang
transform, Fourier transform and wavelet transform in pavement profile
analysis. Vehicle Syst Dyn 2009;47:437–56.
[98] Mohammadzadeh Niaki AH, Afsharnia S. A new passive islanding detection
method and its performance evaluation for multi-DG systems. Electr Power
Syst Res 2014;110:180–7.
[99] Guiliang Y. A distributed generation islanding detection method based on
artificial immune system. In: IEEE transmission and distribution conference
and exhibition: Asia and Pacific; 2005. p. 1–4.
[100] Kar S, Samantaray SR. Data-mining-based intelligent anti-islanding
protection relay for distributed generations. IET Gener Transm Dis
2014;8:629–39.
[101] 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.
[102] Far HG, Rodolakis AJ, Joos G. Synchronous distributed generation islanding
protection using intelligent relays. IEEE Trans Smart Grid 2012;3:1695–703.
[103] 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.
[104] Arachchige LW, Rajapakse A. A pattern recognition approach for detecting
power islands using transient signals; Part I: Design and implementation. In:
IEEE Power & Energy Soc Gen Meeting; 2011. p. 1.
[105] 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.
[106] 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.
[107] 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.
[108] 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 Sust
Energy Rev 2013;27:525–32.
[109] Fayyad Y, Osman A. Neuro-wavelet based islanding detection technique. In:
IEEE electric power and energy conference (EPEC); 2010. p. 1–6.
[110] Brown RA, Lauzon ML, Frayne R. A general description of linear time-
frequency transforms and formulation of a fast, invertible transform that
samples the continuous s-transform spectrum nonredundantly. IEEE Trans
Signal Process 2010;58:281–90.
[111] 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.
[112] Mohanty S, Kishor N, Ray P, Catalao J. Comparative study of advanced signal
processing techniques for islanding detection in a hybrid distributed
generation system. Sust Energy, IEEE Trans Sust Energy 2014:1–10.
[113] Khamis A, Shareef H, Mohamed A, Bizkevelci E. Islanding detection in a
distributed generation integrated power system using phase space technique
and probabilistic neural network. Neurocomputing 2015;148:587–99.
[114] Aziah Khamis HS. An effective islanding detection and classification method
using neuro-phase space technique. World Acad Sci, Eng Technol
2013;7:2067–75.