Renewable and Sustainable Energy Reviews 41 (2015) 495–505

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

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

A critical review of detection and classification of power quality events

Om Prakash Mahela n, Abdul Gafoor Shaik, Neeraj Gupta
Centre for ICT, Indian Institute of Technology, Jodhpur 342011, India

art ic l e i nf o a b s t r a c t

Article history: Requirement of green supply with higher quality has been consumers’ demand around the globe.
Received 17 June 2014 The electrical power system is expected to deliver undistorted sinusoidal rated voltage and current
Received in revised form continuously at rated frequency to the consumers. This paper presents a comprehensive review of signal
18 August 2014
processing and intelligent techniques for automatic classification of the power quality (PQ) events and
Accepted 26 August 2014
Available online 16 September 2014
an effect of noise on detection and classification of disturbances. It is intended to provide a wide
spectrum on the status of detection and classification of PQ disturbances as well as an effect of noise on
Keywords: detection and classification of PQ events to the researchers, designers and engineers working on power
Power system quality. More than 150 research publications on detection and classification techniques of PQ
Feature extraction
disturbances have been critically examined, classified and listed for quick reference.
Power quality
& 2014 Elsevier Ltd. All rights reserved.
PQ disturbance
PQ event classifier
Segmentation

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
2. Power quality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
3. Automatic power quality events’ classifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
4. Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
5. Feature extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
5.1. Fourier transform based methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
5.2. S-transform based methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
5.3. Hilbert Huang transform based methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
5.4. Wavelet transform based methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
5.5. Miscellaneous feature extraction techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
5.6. Comparative study of PQ events’ detection techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
6. Artificial intelligence classification techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
6.1. Support vector machine based classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
6.2. Neural network based classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
6.3. Fuzzy expert system based classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500
6.4. Neuro-fuzzy system based classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500
6.5. Genetic algorithm based classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500
6.6. Miscellaneous classification systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500
6.7. Comparative study of PQ events classification techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
7. Effect of noise on PQ event classifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
8. Future scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
9. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502

n
Corresponding author.
E-mail addresses: [email protected] (O.P. Mahela), [email protected] (A.G. Shaik), [email protected] (N. Gupta).

http://dx.doi.org/10.1016/j.rser.2014.08.070
1364-0321/& 2014 Elsevier Ltd. All rights reserved.
496 O.P. Mahela et al. / Renewable and Sustainable Energy Reviews 41 (2015) 495–505

1. Introduction feature extraction techniques which is sub-classified into Fourier

transform based methods [24–36], S-transform based methods
Considerable changes in a business environment have increased [37–44], Hilbert Huang transform based methods [45–54], wavelet
the use of sensitive electronic components, computers, programmable transform based methods [55–69], and miscellaneous techniques
logic controllers, protection and relaying equipments which have [70–87]. The fourth category [88–164] is on artificial intelligent
increased the power consumptions [1]. Increasing consumer expecta- classification techniques which is further classified into support
tions with the requirement of green supply around the globe, where vector machine based classification [89–102], neural network
integration of renewable energy sources to the distribution grid is the based classification [103–115], fuzzy-expert system based classifi-
focus area of smart grid, electrical power systems are expected to cation [116–129], neuro-fuzzy system based classification [130–
deliver power supply continuously at high quality to the consumers. 136], genetic algorithm based classification [137–145] and mis-
Economy of any country suffers with huge losses when there are cellaneous classification systems [146–164].The fifth and final
voltage or current abnormalities present in the power delivery. Any category [165–171] is on the effect of noise on PQ events’ classifica-
deviation/disturbance manifested in the voltage, current and fre- tion. However some publications include more than one category
quency from the standard rating is treated as a power quality (PQ) and have been classified based on their dominant field.
problem that results in failure or malfunctioning of electrical/electronic This paper is divided into nine sections. Starting with an
equipments [2]. Power quality disturbances increase the risk of black- introduction in Section 1, Section 2 describes power quality and
out; especially because of the failure of interdependencies between Section 3 covers the automatic power quality event classifier. The
sub-networks and associated dynamical propagations. To prevent segmentation of PQ disturbances, techniques for feature extraction
these issues customers are willing to invest in the on-site equipments from PQ disturbances and artificial intelligent classification tech-
to ensure higher level of quality supply such as uninterrupted power niques are covered under Sections 4–6. The effect of noise on PQ
supply (UPS) and stabilizers even though these are very costly [3]. It event classification and future scope are presented in Sections 7 and 8.
shows the importance of power quality towards economic distribution Finally the concluding remark is included in Section 9.
of the energy.
Renewable energy integration and smart transmission systems,
well equipped with modern control equipments, increase the 2. Power quality
applications of nonlinear and electronically switched devices in
distribution systems which exaggerate problems such as harmo- The term power quality (PQ) is generally applied to a wide
nics, flicker, voltage sag/swell, voltage regulation, load unbalancing variety of electromagnetic phenomena occurring within a power
and deviations in phase as well as frequency. Several solid-state system network [2]. The ability of the power systems to deliver
electronic/power-electronic devices have been developed, studied undistorted voltage, current and frequency signals is termed
and proposed to the international scientific community with the as quality of power supply [12]. Unexpected variation of the
goal of improving supplied power quality in last two decades [4]. voltage or current from normal characteristics can damage or shut
This causes increased operational and planning complexity of down the critical electrical equipments designed for specific
electricity supply networks which requires increased attention purpose. Such variations happen in electrical networks with a
for quality of power supply [5]. The transient oscillations are also great frequency due to a competitive environment and continuous
seen in power networks when power electronically controlled change of power supply. In a highly evolved electrical system PQ
capacitors are switched across a node/bus. sensitive demands can be classified as (i) digital economy (such as
The harmonics and power sinusoids modulated by low frequency banking, share market and railways), (ii) continuous process
signals are also observed in electrical networks due to sudden changes manufacturing industries, and (iii) fabrication and essential ser-
in the operating conditions such as load disturbances, network vices. Cost incurred to operate all the above types of loads vary
contingencies, and output of renewable energy sources [6–10]. These from 3 to 120 per kVA per event [12]. This is huge and greatly
events may impose penalties to the consumers in terms of supporting affects economic operation of power industries. To mitigate PQ
ancillary services’ cost such as cost of reactive power, re-dispatch cost issues; customers are also equipped with some back-up instru-
and load curtailment cost to remove disturbing load. Therefore, PQ ments apart from grid supply [13]. According to IEEE standard
events need to be monitored and mitigated to maintain standard of 1159-1995 [14], the PQ disturbances include a wide range of PQ
supply and economic operating conditions. In complex and large phenomena namely transient (impulsive and oscillatory), short
power systems, during operations, a huge amount of measured PQ duration variations (interruption, sag and swell), frequency varia-
events data make analysis difficult for monitoring programs. There- tions, long duration variations (sustained under voltages and
fore, a smart tool and methodology are required for detection and sustained over voltages) and steady state variations (harmonics,
classification of this data for clear and smart understanding to the notch and flicker) with a time scale which ranges from tens
utilities, regulators and customers about the operating requirements of nanoseconds to steady state. Inigo Monedero et al. [15] defined
under sudden change of operating conditions [11]. In this regard, PQ disturbances based on UNE standard in Spain as given in Table 1.
feature extraction and classification are the most important part of the IEEE Std. 1459-2010 [16] includes definitions for measurement of
generalized PQ event classification system where PQ event detection electric power quantities under sinusoidal, non-sinusoidal, balanced,
requires the feature extraction from the disturbances. To make analysis and unbalanced conditions. In [17], authors presented an IEEE Std.
more effective; a set of better features should be available to report the 1459 power magnitude measurement system working as a part of a
disturbance signal efficiently [4]. Thereafter, this set of extracted PQ improvement structure.
features can be used for the classification process.
This paper aims at presenting a comprehensive review on the
topic of power quality events detection and classification techni- 3. Automatic power quality events’ classifier
ques/methodologies. Over 150 publications [1–171] are critically
reviewed and classified into six major categories. The first [1–17] is Before 1990s some manual configurations were used for monitor-
on general power quality and their economics as well as standards. ing and managing the quality of power supply due to lack of signal
The second category [18–23] is on the automatic power quality processing and intelligent techniques’ advancement. The decision was
events’ classifier and general ideas of segmentation and feature based on the operator/expert intuition to maintain quality within
extraction of PQ disturbances. The third category [24–87] includes a certain range. As technology evolved, smart signal processing
 O.P. Mahela et al. / Renewable and Sustainable Energy Reviews 41 (2015) 495–505 497

Table 1
PQ disturbances’ classification [15].

Type Disturbance type Time Range

Min. value Max. value

Frequency Slight deviation 10 s 49.5 Hz 50.5 Hz

Severe deviation 47.0 Hz 52.0 Hz

Voltage Average voltage 10 min 0.85 Un 1.1 Un

Flicker – – 7%
Sag Short 10 ms–1 s 0.1 U 0.9 U
Long 1 s–1 min
Long-time disturbance 4 1 min
Under voltage Short o 3 min 0.99 U
Long 4 3 min
Swell Temporary short 10 ms–1 s 1.1U 1.5 kV
Temporary long 1 s–1 min
Temporary long-time 4 1 min
Over voltage o 10 ms 6 kV

Harmonics and other Harmonics – THD 4 8%

information signals Information signals – Included in other
disturbances

Fig. 1. Block diagram automatic power quality event classifier.

techniques (such as pattern-recognition, data mining, internet-net- include short-term Fourier transform (STFT) and wavelet transform
working, and artificial intelligence) with intelligent instruments (such (WT) [22].
as computers, digital signal processors, mass storage, information and
communications) are adopted time to time. A basic block diagram of
automatic PQ event classifier is given in Fig. 1 [18]. The disturbance 5. Feature extraction
signal is passed through the pre-processing unit having two function
blocks; segmentation and feature extraction. The extracted features of Feature extraction from PQ disturbances is also known as
disturbances are passed through processing unit for classification. detection of disturbances. Extracted features are used to classify
Finally, the post-processing unit gives an actual decision regarding the the PQ events. Thereafter, the classifier's information is used to
type of disturbance. make the final decision through a post-processing unit [18]. The
selection of suitable features of PQ events is extremely important
for classification. Features may directly be extracted from the
4. Segmentation original measurement either from some transformed domain or
from the parameters of signal models [23]. In this context, recent
Segmentation divides the data sequence into stationary and non- developments regarding feature extraction techniques are dis-
stationary parts [19]. Events are the segments in between transition cussed in the subsections as detailed below.
segments. The feature is extracted from the event segments because
the signal is stationary and normally contains the information that is 5.1. Fourier transform based methods
unique enough to distinguish among different types of disturbances.
To capture the disturbance waveform period, a triggering method is The best known technique for frequency domain analysis is
required to get start and end time instant of PQ event [20,21]. The the Fourier transform (FT), where it represents a signal as sum of
current methods used for detecting PQ disturbances are based on sinusoidal terms of different frequencies [24]. FT is suitable for
a point-to-point comparison of adjacent cycle or a point-to-point stationary signals and extracting spectrum at specific frequen-
comparison of the RMS values of the distorted signal with its cies; however it is unable to resolve any temporary information
corresponding pure signal and/or frequency domain transformed data. associated with fluctuations [25]. One variant of FT; the short
The recent methods proposed for this purpose are classified as time Fourier transform (STFT) divides the signal into small
parametric (or model based) and non-parametric (or transform based). segments where each segment can be assumed to be stationary
The parametric methods include techniques such as Kalman filter (KF) [26]. In this regard, STFT determines the sinusoidal frequency and
and auto-regressive (AR) models while non-parametric techniques phase contents of local sections of signal as they change over
498 O.P. Mahela et al. / Renewable and Sustainable Energy Reviews 41 (2015) 495–505

time. Also, it extracts several frames of signals to be analyzed development, multiple HHT variants are demonstrated in the
with a window that moves with time. With a moving window, context of PQ detection and classification.
the relation between the variance of frequency and the time can In [48], authors developed a method based on combination of EMD
be identified [27,28]. It is difficult to analyze non-stationary and Hilbert transform for assessment of power quality events. How-
signals with STFT [29]; however it has been applied to non- ever, an approach for power quality disturbances’ classification using
stationary signals when operating in a fixed window size [30]. HHT and Relevance Vector Machine (RVM) is presented in [49]. In [50],
Discrete STFT is used for time–frequency analysis of non- authors presented the multi-disturbance complex power quality signal
stationary signals. It decomposes the time-varying signals into HHT detection method using the HHT algorithm. In [51], a voltage
time–frequency domain components [31]. Discrete Fourier trans- flicker is analyzed using the orthogonal Hilbert Huang Transform
form (DFT) represents the discrete signals that repeat themselves (OHHT) instead of HHT which can be used in the analysis of nonlinear
in a periodic fashion from negative to positive infinity whereas and non-stationary signals. In [52], authors proposed a novel method
fast Fourier transform (FFT) gives exactly the same result as the based on mathematical morphology and HHT which is used to detect
DFT in much less time [32]. and analyze power quality disturbances. In [53], authors presented the
In [33], authors presented unique features that characterize PQ Hilbert-Huang method with modifications for time–frequency analysis
events and methodologies to extract them from recorded voltage and/ of distorted power quality signals. In [54], authors presented a method
or current waveforms using Fourier and Wavelet Transforms (WT). to extract efficient features of the PQ disturbances using Hilbert
In [34], authors used windowed FFT for power quality assessment. Transform and to classify the disturbance signal using radial basis
The windowed FFT is a time version of the discrete time FT. It was function (RBF) neural networks with more improved methodology.
experimentally proved that WT is better than STFT [29]. The schemes
based on STFT, S-transform, and Kalman filter have been developed 5.4. Wavelet transform based methods
efficiently for detecting PQ events [35,36].
The wavelet transform (WT) is a mathematical tool, much like FT,
that decomposes a signal into different scales with different levels of
5.2. S-transform based methods
resolution by dilating a signal prototype function. The WT is based on
a square-integral function and group theory representation. The WT
The S-transform is a time–frequency tool generated by the
provides a local representation (in both time and frequency) of a given
combination of WT and STFT [37]. It produces a time–frequency
signal; therefore it is suitable for analyzing a signal where time–
representation of a time series. It uniquely combines a frequency-
frequency resolution is needed such as disturbance transition events
dependent resolution that simultaneously localizes the real
in power quality [55,56]. The WT is classified into discrete wavelet
and imaginary spectra. The basic function for the S-transform is
transform (DWT) and continuous wavelet transform (CWT) [57].
Gaussian modulated co-sinusoids [38]. In the case of non-stationary
In [58], authors introduced the use of WT and multi-resolution
disturbances with noisy data; the S-transform provides patterns
signal decomposition as a powerful analysis tool for PQ events.
that closely resemble the disturbance type and thus requires a
A wavelet based method for detecting, localizing, quantifying and
simple classification procedure [28].
classifying short duration PQ disturbances is presented in [59].
In [39], authors proposed a simple and effective method for
In [60], authors introduced a compression technique for power
classification and quantification of ten typical kinds of power
disturbance data via DWT and wavelet packet transform (WPT).
quality disturbances using S-transform. In [40], authors presented
In [61], authors proposed a model of disturbance detection for
a real-time power quality disturbances’ classification by using
harmonics and voltages using wavelet probabilistic network which
a hybrid method based on S-transform and dynamics where the
is a two-layer architecture containing the wavelet layer and the
dynamics is used to reduce run time. In [41], a S-transform based
probabilistic network. Consequently, a novel approach for the PQ
neural network structure is presented for automatic classification of
disturbances’ classification based on the WT and self-organizing
power quality disturbances. The S-transform technique is integrated
learning array system is proposed in [62]. In [63], authors
with neural network model with multi-layer perception to con-
introduced a new perspective for the IEEE standard 1459-2000
struct the classifier. Power quality analysis using discrete orthogonal
definitions using the stationary wavelet transform (SWT) for
S-transform is presented in [42]. Multi-resolution S-transform
defining power components, power factors, and pollution factor.
based fuzzy recognition system for power quality events is pro-
In [64], PQ indices that were recommended in [65] and [66] are
posed in [43]. This is based on a variable width analysis window
redefined in the time frequency domain using WPT. A wavelet
which changes with frequency according to user defined function.
norm entropy based effective feature extraction method for PQ
In [44], authors proposed more suitable fast variants of the discrete
disturbance classification is presented in [67]. In [68], authors
S-transform (FDST) algorithm to accurately extract the time loca-
proposed a WT and S-transform based approach for islanding
lized spectral characteristics of non-stationary signals.
detection and disturbance due to load rejection in the distributed
generation (DG) based hybrid system. In [69], authors introduced
5.3. Hilbert Huang transform based methods the un-decimated wavelet transform to compute power quantities
using complex wavelet coefficients. The important issue related to
The Hilbert Huang transform (HHT), a novel signal processing the use of wavelet method is the choice of suitable wavelet. The
algorithm was proposed in 1998 by Dr. Huang which consists of computational cost increases with an increase in filter length.
two distinct processes [45]. The signal to be analyzed is decom- Table 2 provides characteristics of four commonly used wavelets.
posed using the empirical mode decomposition (EMD) process
into intrinsic mode function (IMF) that have meaningful instanta- 5.5. Miscellaneous feature extraction techniques
neous frequencies and amplitudes. The EMD decomposes the
signal into IMFs in such a way that the IMFs are sorted from the Apart from the above techniques mentioned in Sections 5.1–5.4,
highest frequency to the lowest frequency. Once the signal is some other techniques have played a significant role in PQ detection
decomposed into IMFs, the Hilbert transform can then be applied and classification. De-noising techniques with change point approach
to each IMF giving the instantaneous amplitude and instantaneous for wavelet based power quality monitoring are used and demon-
frequency versus time curve. This combination of EMD process strated in [70]. Parallel computing for time–frequency feature extrac-
and Hilbert transform is known as the HHT [46,47]. In a row of tion of PQ disturbances is presented in [71]. Hybrid wavelet and HT
 O.P. Mahela et al. / Renewable and Sustainable Energy Reviews 41 (2015) 495–505 499

with frequency shifting decomposition for PQ analysis are presented in detect PQ disturbances is given in Table 3. The comparison of main PQ
[72]. Several other techniques such as Gabor–Wigner transform (GT) disturbance analysis methods is detailed in Table 4.
[73], Kalman filter [74,75], Fuzzy-ARTMAP-wavelet network [76], TT
Transform [77], Curve fitting [78], DWT Transform and wavelet
network [79], Hybrid soft computing technique [80], parametric 6. Artificial intelligence classification techniques
spectral estimation method [81], extended Kalman filtering [82], linear
combiners [83], higher order statistics and case based reasoning [84], A broad definition of artificial intelligence (AI) can be the
Adaline [85], digital filters [86], short-time correlation transform [87] automation of activities that are associated with human thinking
have played an important role in PQ event detection and classification such as decision making, problem solving, learning, perception,
in recent years. and reasoning [88]. In this regard recent developments in AI tools
of interest to the electric power community are detailed below.

5.6. Comparative study of PQ events’ detection techniques

6.1. Support vector machine based classification
Comparative study of PQ events’ detection techniques is carried out
The foundation of Support Vector Machine (SVM) has been
based on critical reviews of publications [24–87]. The computational
developed by Vapnik [89], where statistical learning theory being
efficiency of major power quality disturbance detection techniques to
the basis provides a new pattern recognition approach. SVMs are a set
of related supervised learning methods used for classification and
regression. They belong to family of generalized linear combiners [90].
Table 2
In [91], authors presented an identification method for PQ
Wavelet characteristics.
events based on N-1 SVMs’ classifier which is particularly effective
Wavelet Orthogonal Compact Support Filters Symmetry in the automatic classification of voltage disturbances [92]. Con-
name property support width length sequently, Whei-Lin et al. [93] presented one-versus-one approach
based SVM which can process the multiple classifications of PQ
Haar ✓ ✓ 1 2 ✓
Symlets ✓ ✓ 2N-1 2N Near to disturbances. The SVM method is better than optimal time
symmetry frequency representation (OTFR) and is a low complexity event
Daubechies ✓ ✓ 2N-1 2N Near to classifier [94]. An integrated model for recognizing PQ distur-
symmetry bances using wavelet multi-class SVM is presented in [95]. A SVM
Coiflets ✓ ✓ 6N-1 6N No
classifier combined with DWT to recognize the type of power
N-Sample points for entire observation time. system PQ is presented in [96]. The direct acyclic graph SVM
correctly classifies the PQ events with high degree of accuracy and
less training as well as testing times in comparison to other
Table 3
Computational efficiency of PQ detection techniques.
kernel-based learning techniques and ANN based methods [97].
The classification of PQ events based on wavelet transform and
Sr. PQ Percentage efficiency of PQ detection techniques SVM is presented in [98–100]. The number of SVM and the
no. disturbance compactness of different clusters are enhanced using modified
Hilbert Huang S- Discrete wavelet Fast Fourier
immune optimization algorithm in classification of non-stationary
transform [48] transform transform [79] transform
[48] [156] power signals using modified TT-transform and SVM [101]. In
[102], authors presented classification of PQ events using SVM and
1 Sag 100 100 98.67 95 higher order statistical features.
2 Swell 100 100 99.33 98
3 Harmonic 95 100 99.33 100
4 Flicker 100 100 98.67 89 6.2. Neural network based classification
5 Notch 100 83 97.33 –
6 Spike 95 77 – –
7 Transient 98 100 98.67 100 Neural networks (NN) represent the promising new generation
8 (1 þ3) 98 100 98.18 – of information processing systems. They are good at tasks such as
9 (2 þ3) 89 100 98.18 – pattern matching, classification, function approximation, optimi-
10 (1 þ7) – – 96.36 –
zation and data clustering [103]. The classification and function
11 (2 þ7) – – 98.18 –
approximation capabilities of artificial neural networks (ANN)

Table 4
Comparison of main methods of PQ disturbances analysis.

Sr. Method Advantages Disadvantages

no.

1 STFT Successfully used for stationary signals where properties of Not suitable for non-stationary signal as it does not track signal dynamics properly due
signals do not evolve in time. Simple in implementation to limitation of fixed window width
2 HHT Useful in feature extraction of distorted waveform, generates Limited only for narrow band conditions
quadrature signal by which instantaneous amplitude and phase
can be easily evaluated
3 ST Fully convertible from time domain to 2-D frequency translation Based on block processing manner and does not satisfy real-time requirement,
domain and then to Fourier frequency domain incorrect measurement of harmonics due to dependency of frequency window width
on central frequency
4 WT Provide local representation in both time and frequency. Strongly influenced by noise present in the signal, suffering from spectral leakage and
Therefore, suitable where good time–frequency resolution is picket fence effects
required.
5 GT It has high signal to noise ratio and good time–frequency Use limited at high frequencies, computational complexity is directly associated with
resolution sampling frequency
500 O.P. Mahela et al. / Renewable and Sustainable Energy Reviews 41 (2015) 495–505

have been employed in power quality studies, fault, and harmonics decision making process, and fast computation using fuzzy num-
source classification. ber operations [130,131].
In [104], authors proposed and implemented a NN approach to In [132], authors described a cork stopper quality classification
the classification of power system disturbance waveforms. Two system using morphological filtering and contour extraction by
different NN paradigms, the common feed-forward neural network following a feature extraction method and a fuzzy-neural network
(FFNN) and time delay neural network (TDNN) are investigated. In as a classifier. In [133], an approach based on a four step algorithm
[105], a method for automatically detecting, localizing and classifying combining 3-D space referential representation, principal compo-
various types of disturbance wave-shape fault is presented. A NN nent analysis and neuro-fuzzy based automatic classification is
based approach to non-intrusive harmonic source identification is presented. Adaptive-neuro-fuzzy-inference system approach for
proposed in [106]. In [107], authors proposed a novel method based transmission line fault classification and location is presented in
on wavelet and ANN for transmission line fault detection and [134,135]. A method for mitigation of voltage sags with phase
classification using oscillo-graphic data. In [108], some patterns based jumps by UPQC with particle-swarm optimization (PSO) based
on DWT are studied for detection and identification of both low adaptive neuro-fuzzy system is presented in [136].
frequency disturbances like flicker, harmonics, and high frequency
disturbances like transient and sags. A wavelet based artificial neural 6.5. Genetic algorithm based classification
network classifier for recognizing power quality disturbances is
implemented and tested in [109]. In [110], a feature extraction Genetic Algorithm (GA) is a search algorithm based on the
method based on centre clustering is obtained which is used as an mechanics of natural selection and natural genetics. It combines
input to ANN for PQ event classification. PQ events’ classification survival of the fittest among string structures with a structured yet
based on S-transform and probabilistic NN is presented in [111]. randomized information exchange to form a search algorithm with
Classification of PQ events using a balanced neural tree is presented some of the innovative flair of human search [137]. The GA is
in [112]. A dual neural network based methodology to detect and considered to be an excellent intelligent paradigm for optimization
classify single and combined PQ disturbances is proposed in [113]. using a multipoint, probabilistic, random and guided search
The radial basis function (RBF) neural network for recognition and mechanism [138].
classification of PQ events is presented in [114,115]. In [139], authors described an innovative and fuzzy-based adaptive
approach to the metering of power and RMS voltage and current
6.3. Fuzzy expert system based classification employing GA. Power system model validation for PQ assessment
applications using GA is presented in [140]. The GA is introduced as
Fuzzy classification system is based on Mamdani type rules to a powerful tool for monitoring and supervising power system
evaluate the information provided by the linguistic variable inputs disturbances generated due to dynamic performance of power sys-
[116,117]. Fuzzy logic refers to a logic system that generalizes the tems in [141]. In [142], a jumping genes paradigm in the format of
classical two-valued logic for reasoning under uncertainty. It hierarchical GA is proposed for optimizing the power voltage control
is motivated by observing that human reasoning can utilize systems. A new method using enhanced GA for placement of PQ
concepts and knowledge that do not have well defined or sharp monitors is proposed in [143]. An analytic method of power quality
boundaries [118]. A fuzzy expert-system is an expert system that using extension GA and wavelet transform is presented in [144].
uses a collection of fuzzy sets and rules instead of Boolean sets for In [145], authors described the design of a GA that optimizes the
reasoning about data [119]. S-transform for analysis and classification of the perturbations in
In [120], authors proposed the design of a tool to quantify PQ electrical signals.
parameters using wavelets’ and fuzzy sets’ theory. A hybrid technique
for characterizing PQ events using a linear Kalman filter and a fuzzy- 6.6. Miscellaneous classification systems
expert system is presented in [121]. An approach for the detection and
classification of single and combined PQ disturbances using fuzzy logic Detection and classification of PQ events in real-time are
and a particle swarm optimization algorithm is proposed in [122]. In important considerations to electric utilities. The design and
[123], authors presented an approach for the classification of PQ data development of a rule based system for intelligent classification
using decision tree and chemo-tactic differential evolution based fuzzy of PQ disturbances using S-transform features are presented in
clustering. An approach for PQ time series data mining using S- [146,147]. Rule-based and wavelet-multi-resolution decomposi-
transform based fuzzy expert system is presented in [124]. In [125], tion for PQ events classification is presented in [148]. An expert
authors presented an approach for the visual localization, detection
and classification of various non-stationary power signals using a
variety of windowing techniques. In [126], a hybrid scheme using a
Table 5
Fourier linear combiner and a fuzzy expert system for the classification Strength and weakness of AI techniques.
of transient disturbance waveforms in a power system is presented. In
[127], authors proposed a wavelet-based extended fuzzy reasoning Sr. No. Attributes AI techniques
approach to PQ disturbance recognition and identification. An adaptive
NN ANN SVM FL GA ES
fuzzy self-learning technique for detection of abnormal operation of
electrical systems is presented in [128]. A data compression technique 1 Uncertainty tolerance ⋆⋆⋆ ⋆⋆⋆ ⋆⋆⋆⋆ ⋆⋆⋆ ⋆⋆⋆ ⋆⋆
for power waveform using adaptive fuzzy logic is presented in [129]. 2 Imprecision tolerance ⋆⋆⋆ ⋆⋆⋆ ⋆⋆⋆⋆ ⋆⋆⋆ ⋆⋆⋆ ⋆
3 Knowledge representation ⋆ ⋆⋆⋆⋆⋆ ⋆ ⋆⋆⋆ ⋆ ⋆⋆
4 Adaptability ⋆⋆⋆ ⋆⋆⋆ ⋆⋆⋆ ⋆⋆ ⋆⋆⋆ ⋆
6.4. Neuro-fuzzy system based classification
5 Maintainability ⋆⋆⋆ ⋆⋆⋆ ⋆⋆⋆ ⋆⋆ ⋆⋆ ⋆
6 Learning ability ⋆⋆⋆ ⋆⋆⋆⋆ ⋆⋆⋆⋆ ⋆ ⋆⋆⋆ ⋆
Neuro-fuzzy based methods have advantages of handling any 7 Explanation ability ⋆ ⋆⋆ ⋆ ⋆⋆⋆ ⋆ ⋆⋆⋆
kind of information, manage imprecise, partial, and vague or 8 Data mining ⋆⋆⋆ ⋆⋆⋆ ⋆⋆⋆ ⋆⋆ ⋆⋆ ⋆
imperfect information. These methods also have advantages to 9 Generalized Performance ⋆⋆⋆ ⋆⋆⋆ ⋆⋆⋆⋆⋆ ⋆ ⋆ ⋆

resolve conflicts by collaboration and aggregation, self-learning, NN – Neural network, ANN – Artificial neural network, SVM – Support vector
self-organizing and self-tuning capabilities. Further, there is no machine, FL – Fuzzy logic, GA – Genetic algorithm, ES – Expert system; AI
need of prior knowledge of relationship of data, mimic human Technique with higher number of ⋆ is more preferred for specific purpose.
 O.P. Mahela et al. / Renewable and Sustainable Energy Reviews 41 (2015) 495–505 501

Table 6
Comparison of PQ events classification methods.

Sr. Method Advantages Disadvantages

No.

1 NN High classification accuracy for mixed PQ disturbances Less efficient under noisy conditions
2 ANN High accuracy for real time applications and provides mathematical Convergence speed and accuracy depends on the network architecture as
flexibility well as noise in the signal
3 SVM Potential to handle large features, provide stable solution to quadratic Poor classification accuracy when training samples are minimum
optimization, high learning processes
4 FL Accurate in modeling and analyzing complex systems Training set for every case is fixed, hence not suitable for new disturbances
5 GA Accurately classify PQ disturbances generated due to dynamic performance High computational time
of power system and damped sub-harmonic signals
6 ES Can be used with or without limited data Expensive system, slow in execution, it is difficult to draw conclusion if
assumptions and actual situations do not match exactly

system for PQ classification is presented in [149,150]. Several other Table 7

PQ classification techniques such as Warping Classifier [151], Effect of noise on PQ classification.

Digital filtering and mathematical morphology [152], Hardware Sr. no. PQ disturbance Percentage error due to noise
and software architecture [153–155], Hidden Markov models and
vector quantization [156], recurrence quantification analysis [157], ANN SVM
Phasor data records and sequence of events [158], Multi-way
1 Sinusoidal waveform 10.86 9.67
principal component analysis [159], nearest neighbor rule [160],
2 Voltage sag 8.69 5.43
Transient-meter [161], Easy VI program [162], inductive inference 3 Voltage swell 8.69 6.45
approach [163], and fault current limiting high-temperature 4 Notch 4.44 4.54
superconductor cable [164] have played an important role in PQ 5 Transient 6.67 5.49
event classification in past years.

6.7. Comparative study of PQ events classification techniques with a spatial noise-suppression method for wavelet-based PQ mon-
itoring is presented in [171].
Comparative study of PQ events’ classification techniques is A brief comparative study of error introduced due to the
carried out based on critical reviews of publications [88–164]. The presence of noise in classification of PQ disturbances using ANN
strength and weakness of different AI techniques for power quality and SVM is presented in Table 7.
disturbance classification are analyzed and presented in Table 5.
The comparison of different power quality disturbances’ classifica-
tion methods is provided in Table 6.
8. Future scope

The present thrust in PQ research is of real-time analysis;

7. Effect of noise on PQ event classifiers
therefore many researchers have given so many enhancements in
real-time detection and mitigation. However, the PQ analysis
The signals captured by monitoring devices are often accom-
technology still has a way to grow especially for real-time PQ
panied with noise thereby affecting the extraction of important
events. Therefore, there is a need to work on real-time PQ events
features from the signal. Noise has an adverse effect on the
to develop new techniques for classification and mitigation. There
performance of wavelet based event detection, time localization
is also a need to develop a generalized approach for detection and
and classification schemes due to the difficulty of separating noise
classification of single and multiple PQ events. On this basis, it is
and disturbances [165]. The disturbed components of the wave-
possible to improve the study of blackout conditions.
form carry the most important information for detection and
classification. When PQ data is decomposed using wavelet trans-
form most of the disturbance components are reflected at higher
frequency bands which are also occupied by noise. Therefore, even 9. Conclusion
if the magnitude of noise level present is not very high compared
to fundamental component for many PQ disturbances it is com- A critical and comprehensive literature review on the detection
parable to disturbance energy at these bands. Hence, the presence and classification of PQ events in the electrical power system is carried
of noise degrades the detection capability of wavelet based PQ out. This paper presents a literature survey on detection
monitoring system [166]. /feature extraction techniques such as wavelet transform, S-transform,
In this regard, a de-noising technique with a change-point Fourier transform, Hilbert-Huang transform and artificial intelligence
approach for wavelet-based power-quality monitoring is presented techniques for PQ classification such as SVM, ANN, Fuzzy logic, and GA.
in [167]. Consequently, performance improvement of power quality The effect of noise on PQ event classification is also outlined. Finally, at
disturbances’ classification based on a new de-noising technique is the end of paper future scope for research in the field of power quality
presented in [168]. Also recognizing noise influenced PQ events with detection and classification techniques has been presented.
integrated feature extraction and neuro-fuzzy network is presented in According to developed review, it can be concluded that commonly
[169]. In [170], authors proposed a simple yet effective de-noising used techniques for detection of PQ disturbances are WT, ST, FT and
technique using inter- and intra-scale dependencies of wavelet HHT. The comparative study of these techniques will help in selecting
coefficients to de-noise PQ waveform data for enhanced detection the particular technique for specific application. Regarding event
and time localization of PQ disturbances. The de-noising technique classification, trend has been the use of algorithm based on ANN,
502 O.P. Mahela et al. / Renewable and Sustainable Energy Reviews 41 (2015) 495–505

SVM, GA, ES, and FS. The major advantages and disadvantages as well [29] Gu YH, Bollen MHJ. Time–frequency and time-scale domain analysis of
as effectiveness of the particular methods are outlined. voltage disturbances. IEEE Trans Power Deliv 2000;15(October (4)):1279–84.
[30] Wright PS. Short-time fourier transforms and Wigner–Ville distributions
applied to the calibration of power frequency harmonic analyzers. IEEE Trans
Instrum Meas 1999;48(April (2)):475–8.
References [31] Jurado Francisco, Saenz Jose R. Comparison between discrete STFT and
wavelets for the analysis of power quality events. Electr Power Syst Res
[1] Thapar A, Saha TK, Dong ZY. Investigation of power quality categorization 2002;62:183–90.
and simulating its impact on sensitive electronic equipment. IEEE Power [32] Huang Shyh-Jier, Hsieh Cheng-Tao, Huang Ching-Lien. Application of Morlet
Energy Soc Gen Meet 2004;1:528–33. wavelets to supervise power system disturbances. IEEE Trans Power Deliv
[2] Dugan RC, McGranaghan MF, Beaty HW. Electrical power systems quality. 1999;14(January (1)):235–43.
New York: McGraw-Hill; 1996; 1–38. [33] Santoso Surya, Grady W Mack, Powers Edward J, Lamoree Jeff, Bhatt Siddarth
[3] Flore RA. State of the art in the classification of power quality events, an C. Characterization of distribution power quality events with fourier and
overview. In: 10th international conference on harmonics and quality of wavelet transforms. IEEE Trans Power Deliv 2000;15(January (1)):247–54.
power, vol. 1; 2002. p. 17–20. [34] Heydt GT, Fjeld PS, Liu CC, Pierce D, Tu L, Hensley G. Application of the
[4] Brenna Morris, Faranda Roberto, Tironi Enrico. A new proposal for power windowed FFT to electric power quality assessment. IEEE Trans Power Deliv
quality and custom power improvement: open UPQC. IEEE Trans Power Deliv 1999;14(October (4)):1411–6.
2009;24(October (4)):2107–16. [35] Barros J, Perez E. Automatic detection and analysis of voltage events in
[5] Teke Ahmet, Saribulut Litfu, Tumay Mehmet. A novel reference signal power systems. IEEE Trans Instrum Meas 2006;55(October (5)):1487–93.
generation method for power quality improvement of unified power quality [36] Mishra S, Bhende CN, Panigrahi BK. Detection and classification of power
conditioner. IEEE Trans Power Deliv 2011;26(October (4)):2205–14. quality disturbances using S-transform and probabilistic neural network.
[6] Dalai Sovan, Chatterjee Biswendu, Dey Debangshu, Chakravorti Sivaji, Bhat- IEEE Trans Power Deliv 2008;23(January (1)):280–7.
tacharya Kesab. Rough-set-based feature selection and classification for [37] Rodriguez A, Ruiz JE, Aguado J, Lopez JJ, Martin FI, Munoz F. Classification of
power quality sensing device employing correlation techniques. IEEE Sens J power quality disturbances using S-transform and artificial neural networks.
2013;13(February (2)). In: IEEE international conference on power engineering, energy and elec-
[7] Sharon Daniel, Montano Juan-Carlos, Lopez Antonio, Castilla Manuel, Borras trical drives, Torremolinos, Spain; May 2011.
Dolores, Gutierrez Jaime. Power quality factor for networks supplying [38] Dash PK, Panigrahi BK, Panda G. Power quality analysis using S-transform.
unbalanced nonlinear loads. IEEE Trans Instrum Meas 2008;57(June
IEEE Trans Power Deliv 2003;18(April (2)):406–11.
(6)):1268–74.
[39] Zhao Fengzhan, Yang Rengang. Power-quality disturbance recognition using
[8] Biswal B, Dash PK, Mishra S. A hybrid ant colony optimization technique for
S-transform. IEEE Trans Power Deliv 2007;22(April (2)):944–50.
power signal pattern classification. Expert Syst Appl 2011;38:6368–75.
[40] He Shunfan, Li Kaicheng, Zhang Ming. A real-time power quality distur-
[9] Kwan Kian Hoong, So Ping Lam, Chu Yun Chung. An output regulation based
bances classification using hybrid method based on S-transform and
unified power quality conditioner with Kalman filters. IEEE Trans Ind
dynamics. IEEE Trans Instrum Meas 2013;62(September (9)):2465–75.
Electron 2012;59(November (11)):4248–62.
[41] Uyar Murat, Yildirim Selcuk, Gencoglu Muhsin Tunay. An expert system
[10] Wang Xiaoyu, Yong Jing, Xu Wilsun, Freitas Walmir. Practical power quality
based on S-transform and neural network for automatic classification of
charts for motor starting assessment. IEEE Trans Power Deliv 2011;26(April
(2)):799–808. power quality disturbances. Expert Syst Appl 2009;36:5963–75.
[11] Zhang Ming, Li Kaicheng, Hu Yisheng. A real-time classification method of [42] Reddy M Jaya Bharata, Raghupathy Rama Krishnan, Venkatesh KP, Mohanta
power quality disturbances. Electr Power Syst Res 2011;81:660–6. DK. Power quality analysis using discrete orthogonal S-transform. Digit
[12] Darrow K, Hedman B, Bourgeois T, Rosenblum D. The role of distributed Signal Process 2013;23:616–26.
generation in power quality and reliability. Final Report, Prepared for: New [43] Chilukuri MV, Dash PK. Multi-resolution S-transform based fuzzy recogni-
York State Energy Research and Development Authority; 2005. tion system for power quality events. IEEE Trans Power Deliv 2004;19
[13] Kezunovic Mladen, Liao Yuan. A novel software implementation concept for (January (1)):323–30.
power quality study. IEEE Trans Power Deliv 2002;17(April (2)):544–9. [44] Biswal Milan, Dash PK. Detection and characterization of multiple power
[14] IEEE recommended practice for monitoring electric power quality. IEEE quality disturbances with a fast S-transform and decision tree based
Standards Board, IEEE Std. 1159-1995, New York: IEEE, Inc.; June 1995. classifier. Digit Signal Process 2013;23:1071–83.
[15] Monedero Inigo, Leon Carlos, Ropero Jorge, Garcia Antonio, Manuel Jose, [45] Yang Liang, Yu Jianming, Lai Yongbin. Disturbance source identification of
Montano Juan C. Classification of electrical disturbances in real time using voltage sags based on Hilbert Huang transform. In: IEEE power and energy
neural networks. IEEE Trans Power Deliv 2007;22(7):1288–96. conference, Chengdu, Asia-Pacific; 2010.
[16] Standard definitions for the measurement of electric power quantities under [46] Rilling G, Flandrin P, Goncalves P. On empirical mode decomposition and its
sinusoidal, non-sinusoidal, balanced, or unbalanced conditions. Revision of algorithm. In: IEEE-EURASIP workshop on nonlinear signal image process,
IEEE Std. 1459-2000, IEEE Standard 1459-2010; March 2010. NSIP-03, Grado, Italy; June 2013.
[17] Alfonso-Gil JoseCarlos, Orts-Grau Salvador, Munoz-Galeano Nicolas, Gimeno- [47] Afroni Mohmmad Jasa, Sutanto Danny, Stirling David. Analysis of non-
Sales Francisco J, Segui-Chilet Salvador. Measurement system for a power stationary power-quality waveforms using iterative Hilbert Huang transform
quality improvement structure based on IEEE Std. 1459. IEEE Trans Instrum and SAX algorithm. IEEE Trans Power Deliv 2013;28(October (4)):2134–44.
Meas 2013;62(August (12)):3177–88. [48] Shukla Stuti, Mishra S, Singh Bhim. Empirical-mode decomposition with
[18] Saxena D, Verma KS, Singh SN. Power quality event classification: an Hilbert transform for power quality assessment. IEEE Trans Power Deliv
overview and key issues. Int J Eng Sci Technol 2010;2(3):186–99. 2009;24(October (4)):2159–65.
[19] Barros Julio, Diego Ramon I, De Apraiz Matilde. Applications of wavelets in [49] Hafiz Faeza, Chowdhury A Hasib. An approach for classification of power
electric power quality: voltage events. Electr Power Syst Res 2012;88:130–6. quality disturbances based on Hilbert Huang Transform and relevance vector
[20] Bollen MHJ, Gu IYH. Signal processing of power quality disturbances. New machine. In: 7th IEEE international conference on electrical and computer
York, USA: IEEE Press; 2006. engineering, Dhaka, Bangladesh; December 2012.
[21] Gunal Serkan, Gerek Omer Nezik, Ece Dogan Gokhan, Edizkan Rifat. The [50] Zhan Wang, Xiangjun Zeng, Xiaoxi Hu, Jingying Hu. The multi-disturbance
search for optimal feature set in power quality event classification. Expert complex power quality signal HHT detection technique. In: IEEE PES
Syst Appl 2009;36:10266–73.
conference on innovative smart grid technologies, Tianjin; May 2012.
[22] Ribeiro Moises V, Romano Joao MT, Duque Carlos A. An improved method for
[51] Onal Yasemin, Turhal U Cigdem. The orthogonal Hilbert Huang transform
signal processing and compression in power quality evaluation. IEEE Trans
application in voltage flicker analysis. In: 4th IEEE international conference
Power Deliv 2004;19(April (2)):464–71.
on power engineering, energy and electrical drives, Istanbul, Turkey; May
[23] Ji TY, Wu QH, Jiang L, Tang WH. Disturbance detection, location and
2013.
classification in phase space. IET Gener Transm Distrib 2011;5(2):257–65.
[52] Huang Yong, Liu Yongqiang, Hong Zhiping. Detection and location of power
[24] Karimi M, Mokhtari H, Iravani MR. Wavelet based on-line disturbance
quality disturbances based on mathematical morphology and Hilbert-Huang
detection for power quality application. IEEE Trans Power Deliv 2000;15
(October (4)):1212–20. transform. In: 9th IEEE international conference on electronic measurement
[25] Axelberg Peter GV, Gu IYH, Bollen MHJ. Support vector machine for and instruments, Beijing, China; August 2009.
classification of voltage disturbances. IEEE Trans Power Deliv 2007;22(July [53] Senoroy Nilanjan, Suryanarayanan Siddharth, Ribeiro Paulo F. An improved
(3)):1297–303. Hilbert-Huang method for analysis of time-varying waveforms in power
[26] Granados-Lieberman D, Romero-Troncoso RJ, Osornio-Rios RA, Garcia-Perez quality. IEEE Trans Power Syst 2007;22(November (4)).
A, Cabal-Yepez E. Techniques and methodologies for power quality analysis [54] Jayasree T, Devraj D, Sukanesh R. Power quality disturbance classification
and disturbances classification in power system: a review. IET Gener Transm using Hilbert transform and RBF networks. Neurocomputing 2010;73:
Distrib 2011;5(4):519–29. 1451–6.
[27] Dash PK, Panigrahi BK, Sahoo DK, Panda G. Power quality disturbance data [55] Santoso Surya, Powers Edward J, Mack Grady W, Hofmann Peter. Power
compression, detection and classification using integrated spline wavelet quality assessment via wavelet transform analysis. IEEE Trans Power Deliv
and S-transform. IEEE Trans Power Deliv 2003;18(April (2)):595–600. 1996;11(April (2)):924–30.
[28] Lee IWC, Dash PK. S-transform based intelligent system for classification of [56] Santoso S, Powers EJ, Grady WM. Power quality disturbance data compres-
power quality disturbance signals. IEEE Trans Ind Electron 2003;50(August sion using wavelet transform methods. IEEE Trans Power Deliv 1997;12(July
(4)):800–5. (3)):1250–7.
 O.P. Mahela et al. / Renewable and Sustainable Energy Reviews 41 (2015) 495–505 503

[57] Angrisani L, Daponte P, Apuzzo MD, Testa A. A measurement method based [85] Abdel-Galil TK, El-Saadany EF, Salama MMA. Power quality event detection
on the wavelet transform for power quality analysis. IEEE Trans Power Deliv using adaline. Electr Power Syst Res 2003;64:137–44.
1998;13(October (4)):990–8. [86] Chen Z, Urwin P. Power quality detection and classification using digital
[58] Gaouda AM, Salama MMA, Sultan MR, Chikhani AY. Power quality detection filters. In: IEEE porto power technical conference, Porto, Portugal; September
and classification using wavelet multi-resolution signal decomposition. IEEE 2001.
Trans Power Deliv 1999;14(October (4)):1469–75. [87] Zhang H, Liu P, Malik OP. Detection and classification of power quality
[59] Xiangxun Chen. Wavelet based detection, localization, quantification and disturbances in noisy conditions. IEE Proc Gener Transm Distrib 2003;150
classification of short duration power quality disturbances. IEEE Power Eng (September (5)):567–72.
Soc Winter Meet 2002;2:931–6. [88] Anis Ibrahim Wael R, Morcos Medhat M. Artificial intelligence and advanced
[60] Hamid Effrina Yanti, Kawasaki Zen-Ichiro. Wavelet based data compression mathematical tools for power quality applications: a survey. IEEE Trans
of power system disturbances using the minimum description length Power Deliv 2002;17(April (2)):668–73.
criterion. IEEE Trans Power Deliv 2002;17(April (2)):462–6. [89] Vapnik V. The nature of statistical learning theory. New York: Springer-
[61] Lin CH, Tsao MC. Power quality detection with classification enhanceable Verlag; 1995.
wavelet probabilistic network in a power system. IEE Proc Gener Transm [90] Khasnobish Anwesh, Bhattacharya Saugat, Konar Amit, Tibarewala DN, Nagar
Distrib 2005;152(November (6)):969–76. Atulya K. A two-fold classification for composite decision about localized
[62] He Haibo, Starzyk Janusz A. A self organizing learning array system for power arm movement from EEG by SVM and QDA techniques. In: IEEE international
quality classification based on wavelet transform. IEEE Trans Power Deliv joint conference on neural networks, San Jose, California, USA; 2011.
2006;21(January (1)):286–95. [91] Ganyun Lv, Wang Xiaodong, Zhang Haoran, Zhang Changjiang. PQ distur-
[63] Morsi Walid G, El-Hawary ME. A new perspective for the IEEE standard bances identification based on SVMs classifier. In: Proceedings of IEEE
1459–2000 via stationary wavelet transform in the presence of non international conference on neural networks and brain, vol. 1; 2005. p.
stationary power quality disturbance. IEEE Trans Power Deliv 2008;23 222–6.
(October (4)):2356–65. [92] Janik Przemyslaw, Lobos Tadeusz. Automated classification of power quality
[64] Morsi Walid G, El-Hawary ME. Wavelet packet transform based power disturbances using SVM and RBF networks. IEEE Trans Power Deliv 2006;21
quality indices for balanced and unbalanced three phase systems under (July (3)):1663–9.
stationary or non-stationary operating conditions. IEEE Trans Power Deliv [93] Whei-Lin, Wu Chien-Hsien, Lin Chia-Hung, Cheng Fu-Sheng. Classification of
2009;24(October (4)):2300–10. multiple power quality disturbances using support vector machine and one-
[65] IEEE recommended practice and requirement for harmonic control in versus-one approach. In: IEEE international conference on power system
electric power systems. IEEE Std. 519-1992; April 1993. technology, Chongqing; 2006.
[66] Definitions for the measurement of electric quantities under sinusoidal, non [94] Cerqueira Augusto Santiago, Ferreira Danton Diego, Ribeiro Moises Vidal,
sinusoidal, balanced, or unbalanced conditions. IEEE Std. 1459-2000; January Duque Carlos Augusto. Power quality events recognition using a SVM based
2000. method. Electr Power Syst Res 2008;78:1546–52.
[67] Uyar Murat, Yildirim Selcuk, Gencoglu Muhsin Tunay. An effective wavelet [95] Whei-Lin, Wu Chien-Hsien, Lin Chia-Hung, Cheng Fu-Sheng. Detection and
based feature extraction method for classification of power quality distur- classification of multiple power quality disturbances with wavelet multiclass
bance signals. Electr Power Syst Res 2008;78:1747–55. SVM. IEEE Trans Power Deliv October 2008;23(4):2575–82.
[68] Ray Prakash K, Mohanty Soumya R, Kishor Nand. Disturbance detection [96] Ekici Sami. Classification of power system disturbances using support vector
in grid connected distributed generation system using wavelet and machines. Expert Syst Appl 2009;36:9859–68.
S-transform. Electr Power Syst Res 2011;81:805–19. [97] Panigrahi BK, Dash PK, Reddy JBV. Hybrid signal processing and machine
[69] Zafar T, Morsi WG. Power quality and the un-decimated wavelet transform: intelligence techniques for detection, quantification and classification of
an analytic approach for time varying disturbances. Electr Power Syst Res power quality disturbances. Eng Appl Artif Intell 2009;22:442–54.
2013;96:201–10. [98] Eristi Huseyin, Demir Yakup. A new algorithm for automatic classification of
[70] Dwivedi UD, Singh SN. Denoising techniques with change-point approach power quality events based on wavelet transform and SVM. Expert Syst Appl
for wavelet based power quality monitoring. IEEE Trans Power Deliv 2009;24 2010;37:4094–102.
(July (3)):1719–27. [99] Eristi Huseyin, Ucar Aysegul, Demir Yakup. Wavelet based feature extraction
[71] Krishna Brahmadesam V, Baskaran Kaliaperumal. Parallel computing for and selection for classification of power system disturbances using support
efficient time frequency feature extraction of power quality disturbances. IET vector machines. Electr Power Syst Res 2010;80:743–52.
Signal Process 2013;7(4):312–26. [100] Moravej Z, Abdoos AA, Pazoki M. Detection and classification of power
[72] Tse Norman CF, Chan John YC, Lau Wing-Hong, Lai Loi Lei. Hybrid wavelet quality disturbances using wavelet transform and support vector machine.
and Hilbert transform with frequency-shifting decomposition for power Electr Power Compon Syst 2010;38:182–96.
quality analysis. IEEE Trans Instrum Meas 2012;61(December (12)):3225–33. [101] Biswal B, Biswal MK, Dash PK, Mishra S. Power quality event characterization
[73] Cho Soo-Hwan, Jang Gilsoo, Kwon Sae-Hyuk. Time–frequency analysis of using support vector machine and optimization using advanced immune
power quality disturbances via Gabor–Wigner transform. IEEE Trans Power algorithm. Neurocomputing 2013;103:75–86.
Deliv 2010;25(January (1)):494–9. [102] Palomares-Salas JC, Aguera-Perez A, De La Rosa JJG. Support vector machine
[74] Styvaktakis E, Bollen MHJ, Gu IYH. Expert system for classification and for power quality disturbances classification using higher statistical features.
analysis of power system events. IEEE Trans Power Deliv 2002;17(April In: Proceedings of IEEE 7th international conference workshop on compat-
(2)):423–8. ibility and power electronics, vol. 1; 2011. p. 6–10.
[75] Macias JAR, Exposito A Gomez. Self-tuning of Kalman filters for harmonic [103] Saini Manish Kumar, Kapoor Rajiv. Classification of power quality events—a
computation. IEEE Trans Power Deliv 2006;21(January (1)):501–3. review. Electr Power Syst Res 2012;43:11–9.
[76] Decanini Jose GMS, Tonelli-Neto Mauro S, Malange Fernando CV, Minussi [104] Ghosh Atish K, Lubkeman David L. The classification of power system
Carlos R. Detection and classification of voltage disturbances using a fuzzy- disturbance waveforms using a neural network approach. IEEE Trans Power
ARTMAP-wavelet network. Electr Power Syst Res 2011;81:2057–65. Deliv 1995;10(January (1)):109–15.
[77] Suja S, Jerome Jovitha. Pattern recognition of power signal disturbances [105] Dilokratanatrakool Choosak, Na Ayudhya Piyasawat Navaratana, Chayavanich
using S-transform and TT transform. Electr Power Energy Syst Tasanee, Prapanavarat Cherdchai. Automatic detection-localization of point
2010;32:37–53. on waveform and classification of power quality disturbance wave-shape
[78] Santoso S, Grady WM, Powers EJ, Lamoree J, Bhatt SC. Characterization of fault using wavelet and neural network. In: Proceedings of 2003 IEEE
distribution power quality events with Fourier and wavelet transforms. IEEE international conference on robotics, intelligent systems and signal proces-
Trans Power Deliv 2000;15(January (1)):247–54. sing, vol. 1; 2003. p. 142–7.
[79] Masoum MAS, Jamali S, Ghaffarzadeh N. Detection and classification of [106] Srinivasan D, Ng WS, Liew AC. Neural network based signature recognition
power quality disturbances using discrete wavelet transform and wavelet for harmonic source identification. IEEE Trans Power Deliv 2006;21(January
networks. IET Sci Meas Technol 2010;4(4):193–205. (1)):398–405.
[80] Manimala K, Selvi K, Ahila R. Hybrid soft computing techniques for feature [107] Silva KM, Souza BA, Brito NSD. Fault detection and classification in transmis-
selection and parameter optimization in power quality data mining. Appl sion lines based on wavelet transform and ANN. IEEE Trans Power Deliv
Soft Comput 2011;11:5485–97. 2006;21(October (4)):2058–63.
[81] Yilmaz Ahmet S, Alkan Ahmet, Asyali Musa H. Applications of parametric [108] Cesar Duarte G, Valdomiro Vega G, Gabriel Ordonez P. Automatic power
spectral estimation methods on detection of power system harmonics. Electr quality disturbances detection and classification based on discrete wavelet
Power Syst Res 2008;78:683–93. transform and artificial intelligence. In: IEEE transmission and distribution
[82] Perez E, Barros J. An extended Kalman filtering approach for detection and conference and exposition, Caracas, Latin America; 2006.
analysis of voltage dips in power systems. Electr Power Syst Res [109] Chandel AK, Guleria G, Chandel R. Classification of power quality problems
2008;78:618–25. using wavelet based artificial neural networks. In: IEEE transmission and
[83] Dash PK, Liew AC, Salama MMA, Mishra BR, Jena RK. A new approach to distribution conference and exposition, Chicago; April 2008.
identification of transient power quality problems using liner combiners. [110] Talaat Nermeen, Ilic Marija. ANNs based on subtractive cluster feature for
Electr Power Syst Res 1999;51:1–11. classifying power quality. In: IEEE 40th North American power symposium,
[84] De La Rosa Juan Jose Gonzalez, Aguera-Perez Agustin, Palomares-Salas Jose Calgary; September 2008.
Carlos, Sierra-Fernandez Jose Maria, Moreno-Munoz Antonio. A novel virtual [111] Huang Nantian, Xu Dianguo, Liu Xiaosheng, Lin Lin. Power quality distur-
instrument for power quality surveillance based in higher-order statistics bances classification based on S-transform and probabilistic neural network.
and case-based reasoning, measurement, vol. 45; 2012. p. 1824–35. Neurocomputing 2012;98:12–23.
504 O.P. Mahela et al. / Renewable and Sustainable Energy Reviews 41 (2015) 495–505

[112] Biswal B, Biswal M, Mishra S, Jalaja R. Automatic classification of power [141] EL-Naggar Khaled M, AL-Hasawi Wael M. A genetic based algorithm
quality events using balanced neural tree. IEEE Trans Ind Electron 2014;61 for measurement of power system disturbances. Electr Power Syst Res
(January (1)):521–30. 2006;76:808–14.
[113] Valtierra-Rodriguez Martin, de Jesus Romero-Troncoso Rene, Osornio-Rios [142] Ma HM, Ng Kai-Tat, Man Kim F. Multiobjective coordinated power voltage
Roque Alfredo, Garcia-Perez Arturo. Detection and classification of single and control using jumping genes paradigm. IEEE Trans Ind Electron 2008;55
combined power quality disturbances using neural networks. IEEE Trans Ind (November (11)):4075–84.
Electron 2014;61(May (5)):2473–82. [143] Hong YY, Chen YY. Placement of power quality monitors using enhanced
[114] Liao Chiung-Chou. Enhanced RBF network for recognizing noise-riding power genetic algorithm and wavelet transform. IET Gener Transm Distrib 2011;5
quality events. IEEE Trans Instrum Meas 2010;59(June (6)):1550–61. (4):461–6.
[115] Jayasree T, Devaraj D, Sukanes R. Power quality disturbance classification [144] Wang Meng-Hui, Tseng Yi-Feng. A novel method of power quality using
using S-transform and radial basis network. Appl Artif Intell 2009;23:680–93. extension genetic algorithm and wavelet transform. Expert Syst Appl
[116] Chacon MI, Duran JL, Santiesteban LA. A wavelet-Fuzzy logic based system to 2011;38:12491–6.
detect and identify electric power disturbances. In: Proceedings of 2007 IEEE [145] Sanchez Pedro, Montoya Francisco G, Manzano-Agugliaro Francisco, Gil
symposium on computational intelligence in image and signal processing. Consolacion. Genetic algorithm for S-transform optimization in the analysis
p. 52–7. and classification of electrical signal perturbations. Expert Syst Appl
[117] Bizjak Boris, Planinsic Peter. Classification of power disturbances using fuzzy 2013;40:6766–77.
logic. In: Proceedings of 2006 IEEE international conference on power [146] Rodriguez A, Aguado JA, Martin F, Lopez JJ, Munoz F, Ruiz JE. Rule based
electronics and motion control, 2006. p. 1356–60. classification of power quality disturbances using S-transform. Electr Power
[118] Yen J, Langari R. Fuzzy logic: intelligence, control, and information. Engle- Syst Res 2012;86:113–21.
wood Cliffs, NJ: Prentice-Hall; 1987. [147] Salen Mohammad E, Mohamed Azah, Samad Salina Abdul. Rule based system
[119] Liao Yuan, Lee Jong-Beom. A fuzzy-expert system for classifying power for power quality disturbance classification incorporating S-transform fea-
quality disturbances. Electr Power Energy Syst 2004;26:199–205. tures. Expert Syst Appl 2010;37:3229–35.
[120] Meher Saroj K, Pradhan Ashok K. Fuzzy classifiers for power quality events [148] Zheng Gang, Shi Mei-Xiang, Liu Ding, Yao Jian, Miao Zhu-Mei. Power quality
analysis. Electr Power Syst Res 2010;80:71–6. disturbance classification based on rule-based and wavelet-multi-resolution
[121] Abdelsalam Abdelazeem A, Eldesouky Azza A, Sallam Abdelhay A. Character- decomposition. In: Proceedings of 2002 IEEE international conference on
ization of power quality disturbances using hybrid technique of linear machine learning and cybernetics, Beijing. p. 2137–41.
Kalman filter and fuzzy-expert system. Electr Power Syst Res 2012;83:41–50. [149] Ibne Reaz Mamum Bin, Choong Florence, Shahiman Sulaiman Mohd, Mohd-
[122] Hooshmand R, Enshaee A. Detection and classification of single and com- Yasin Faisal, Kamada Masaru. Expert system for power quality disturbance
bined power quality disturbances using fuzzy systems oriented by particle classifier. IEEE Trans Power Deliv 2007;22(October (3)):1979–88.
swarm optimization algorithm. Electr Power Syst Res 2010;80:1552–61. [150] Yu Jianming, Wang Leilei, Zhou Bo, Tian Wenbo. An expert system based on
[123] Biswal B, Behera HS, Bisoi R, Dash PK. Classification of power quality data S-transform for classification of voltage dips. In: Proceedings of 2011 IEEE
using decision tree and chemo-tactic differential evolution based fuzzy international conference on artificial intelligence, management science and
clustering. Swarm Evol Comput 2012;4:12–24. electronic commerce, Deng Leng, 2011. p. 3722–35.
[124] Behera HS, Dash PK, Biswal B. Power quality time series data mining using S- [151] Youssef AM, Abdel-Galil TK, El-Saadany EF, Salama MMA. Disturbance
transform and fuzzy expert system. Appl Soft Comput 2010;10:945–55. classification utilizing dynamic time warping classifier. IEEE Trans Power
[125] Biswal Birendra, Dash PK, Panigrahi BK. Power quality disturbance classifica- Deliv 2004;19(January (1)):272–8.
tion using fuzzy C-means algorithm and adaptive particle swarm optimiza- [152] Radil Tomas, Ramos Pedro M, Janeiro Fernando M, Cruz Serra A. PQ
tion. IEEE Trans Ind Electron 2009;56(January (1)):212–20. monitoring system for real-time detection and classification of disturbances
[126] Dash PK, Mishra S, Salam MMA, Liew AC. Classification of power system in a single-phase power system. IEEE Trans Instrum Meas 2008;57(August
disturbances using a fuzzy expert system and a Fourier linear combiner. IEEE (8)):1725–33.
Trans Power Deliv 2000;15(April (2)):472–7. [153] De Yong D, Reineri C, Magnago F. Educational software for power quality
[127] Zhu TX, Tso SK, Lo KL. Wavelet-based fuzzy reasoning approach to power- analysis. IEEE Latin Am Trans 2013;11(February (1)).
quality disturbance recognition. IEEE Trans Power Deliv 2004;19(October [154] Lima Ricardo, Quiroga Damian, Reineri Claudio, Magnago Fernando. Hard-
(4)):1928–35. ware and software architecture for power quality analysis. Comput Electr Eng
[128] Anis Wael R, Morcos Medhat M. An adaptive self-learning technique for 2008;34:520–30.
prediction of abnormal operation of electrical systems. IEEE Trans Power [155] Salem Mohammed E., Mohmed Azah, Samad Salina Abdul, Yahya Iskandar.
Deliv 2006;21(October (4)):1770–7. Software tool for real time power quality disturbance analysis and classifica-
[129] Anis Wael R, Morcos Medhat M. Novel data compression technique for power tion. In: 5th IEEE student conference on research and development, Selangor,
waveforms using adaptive fuzzy logic. IEEE Trans Power Deliv 2005;20 Malaysia; December 2007.
(October (3)):2136–43. [156] Abdel-Galil TK, El-Saadany EF, Youssef AM, Salama MMA. Disturbance
[130] Ajil Kottayil S, Thapliyal Pradip Kumar, Shukla Munn V, Pal Pradip K, Joshi classification using Hidden Markov models and vector quantization. IEEE
Prakash C, Navalgund Ranganath R. A new technique for temperature and Trans Power Deliv 2005;20(July (3)):2129–35.
humidity profile retrieval from infrared-sounder observations using the [157] Wang Xi, Bi Gui-hong, Chen Shi-long, Zu Zhe. A method to analyze power
adaptive neuro-fuzzy inference system. IEEE Trans Geosci Remote Sens system quality disturbing signal based on recurrence quantification analysis.
2010;48(April (4)):1650–9. Proc Eng 2011;15:4115–21.
[131] Mizutani Eiji, Nishio Kenichi. Multi-illuminant color reproduction for elec- [158] Moreto Miguel, Rolim Jacqueline G. Using phasor data records and sequence
tronic cameras via CANFIS neuro-fuzzy modular network device character- of events to automate the classification of disturbances of power generating
ization. IEEE Trans Neural Netw 2002;13(July (4)):1009–22. units. Electr Power Syst Res 2011;81:1266–73.
[132] Chang Joongho, Han Gunhee, Valverde Jose M, Griswold Norman C, Francisco [159] Khoisravi Abbas, Melendez Joaquim, Colomer Joan. Classification of sags
Duque-Carrilo J, Sanchez-Sinencio Edgar. Cork quality classification system gathered in distribution substations based on multiway principal component
using a unified image processing and fuzzy-neural network methodology. analysis. Electr Power Syst Res 2009;79:144–51.
IEEE Trans Neural Netw 1997;8(July (4)):964–73. [160] Gaouda AM, Kanoun SH, Salama MMA. On-line disturbance classification
[133] Frenao Pires V, Amaral Tito G, Martins JF. Power quality disturbances using nearest neighbor rule. Electr Power Syst Res 2001;57:1–8.
classification using the 3-D space representation and PCA based neuro- [161] Daponte P, Di Penta M, Mercurio G. Transientmeter: a distributed measure-
fuzzy approach. Expert Syst Appl 2011;38:11911–7. ment system for power quality monitoring. IEEE Trans Power Deliv 2004;19
[134] Reddy MJ, Mohanta DK. Adaptive-neuro-fuzzy inference system approach for (April (2)):2129–35.
transmission line fault classification and location incorporating effects of [162] Ferrero Alessandro, Salicone Simona. An easy VI program to detect transient
power swings. IET Gener Transm Distrib 2008;2(2):235–44. disturbances in supply voltage. IEEE Trans Instrum Meas 2005;54(August
[135] Reddy M Jaya Bharata, Mohanta Dusmanta Kumar. Performance evaluation of (4)):1725–33.
an adaptive-network-based fuzzy inference system approach for location of [163] Abdel-Galil TK, Kamel M, Youssef AM, El-Saadany EF, Salama MMA. Power
faults on transmission lines using Monte Carlo simulation. IEEE Trans Fuzzy quality disturbance classification using the inductive inference approach.
Syst 2008;16(August (4)). IEEE Trans Power Deliv 2004;19(October (4)):1812–8.
[136] Siva Kumar G, Kalyan Kumar B, Mishra Mahesh K. Mitigation of voltage sags [164] Kim Hee Jin, Shim Jae Woong, Sim Kideok, Hur Kyeon. Assessment of
with phase jumps by UPQC with PSO-based ANFIS. IEEE Trans Power Deliv improved power quality due to fault current limiting HTS cable. IEEE Trans
2011;26(October (4)):2761–72. Appl Superconduct 2013;23(June (3)).
[137] Goldberg DE. Genetic algorithm in search, optimization and machine learn- [165] Hua Liu, Baoqun Zhao, Guangjian Wang. Application of wavelet network for
ing. Reading, MA: Addisson-Wesley; 1989. automatic power quality disturbance recognition in distribution power
[138] Levitin G, Kalyuzhny A, Shenkman A, Chertkov M. Optimal capacitor alloca- system. In: Proceedings of IEEE 26th chinese control conference, Zhangjiajie,
tion in distribution systems using genetic algorithm and a fast energy loss Hunan, China; 2007. p. 254–8.
computation technique. IEEE Trans Power Deliv 2000;15(October (2)):623–8. [166] Panigrahi BK, Sinha SK. Detection and classification of non-stationary power
[139] Kung Chih-Hsein, Devaney Michael J, Huang Chung-Ming, Kung Chih-Ming. disturbances in noisy conditions. In: IEEE international conference on power
Fuzzy-based adaptive digital power metering using a genetic algorithm. IEEE electronics, drives and energy systems, New Delhi, India; 2006.
Trans Instrum Meas 1998;47(February (1)):183–8. [167] Dwivedi UD, Singh SN. Denoising techniques with change-point approach for
[140] El-Zonkoly AM. Power system model validation for power quality assessment wavelet-based power-quality monitoring. IEEE Trans Power Deliv 2009;24
applications using genetic algorithm. Power Syst Appl 2005;29:941–4. (July (3)):1719–27.
 O.P. Mahela et al. / Renewable and Sustainable Energy Reviews 41 (2015) 495–505 505

[168] Hu Wei Bing, Li Kai Cheng, Zhao Dang Jun, Xie Bing Ruo. Performance [170] Dwivedi UD, Singh SN. Enhanced detection of power-quality events using
improvement of power quality disturbance classification based on a new de- intra and inter-scale dependencies of wavelet coefficients. IEEE Trans Power
noising technique. In: Proceedings of IEEE international conference on Deliv 2010;24(October–January (1)):358–66.
electrical machines and systems; October 2007. p. 1806–10. [171] Liao Chiung-Chou, Yang Hong-Tzer, Chang Hsuesh-Hsien. Denoising
[169] Liao Chiung-Chou, Yang Hong-Tzer. Recognizing noise-influenced power technique with a spatial noise-suppression method for wavelet-based power
quality events with integrated feature extraction and neuro-fuzzy network. quality monitoring. IEEE Trans Power Deliv 2011;60(July (6)):1986–96.
IEEE Trans Power Deliv 2009;24(October (40)):2132–41.