Classification of power quality events – A review Manish Kumar Saini b,⇑ , Rajiv Kapoor a a Electronics & Communication Engineering Department, Delhi Technological University, New Delhi, India b Electrical Engineering Department, Deenbandhu Chhotu Ram University of Science & Technology, Sonepat, India article info Article history: Received 28 July 2011 Received in revised form 25 April 2012 Accepted 29 April 2012 Available online 17 June 2012 Keywords: Power quality Signal processing tools Intelligence techniques abstract Power quality (PQ) interest has increasingly evolved over the past decade. The paper surveys the appli- cation of signal processing, intelligent techniques and optimization techniques in PQ analysis. This paper carries out a comprehensive review of articles that involves a comprehensive study of signal processing techniques used for PQ analysis. Within this context intelligent techniques such as fuzzy logic, neural net- work and genetic algorithm as well as their fusion are reviewed. Tabular presentation (i.e. highlighting the important techniques) has also been provided for comprehensive study. Although this review cannot be collectively exhaustive, it may be considered as a valuable guide for researchers who are interested in the domain of PQ and wish to explore the opportunities offered by these techniques for further improve- ment in the field of PQ. Crown Copyright Ó 2012 Published by Elsevier Ltd. All rights reserved. 1. Introduction In the industrialized world, electric power systems have become polluted with unwanted variations in the voltage and current signal. PQ issues [1] are primarily due to continually increasing sources of disturbances that occur in interconnected power grids, which contain large numbers of power sources, trans- mission lines, transformers and loads. In addition, such systems are exposed to environmental disturbances like lighting strikes. Fur- thermore, nonlinear power electronic loads such as converter dri- ven equipment have become increasingly common in power system. Poor quality [2,3] is attributed due to the various power line disturbances. In brief, PQ problems can cause system equip- ment malfunction; computer data loss and memory malfunction of sensitive loads such as computer, programmable logic controller controls, protection and relaying equipment; and erratic operation of electronic controls [4]. Therefore, it is necessary to monitor these disturbances. Continuous monitoring is required because of the increasing demand of clean power as suggested in Refs. [5–7] and monitoring standards are also given in Ref. [8]. Since, disturbances occur in the order of microseconds, a captured event recorded using monitoring system produces mega- bytes of data. As a result, the volume of the recorded data increases significantly, necessitating the development of an efficient tech- nique to compress the data volume. Monitoring has significant implications in the area of PQ [9]. The volume of the data to be recorded and examined is prohibitively large, if all the waveform is to be saved into the instrument or a personal computer. With the advancements in PQ monitoring equipments, data compression has received great attention from those involved in the field [10]. In many countries, high-tech manufacturers often concentrate in industry parks, therefore any PQ events in the utility grid can affect a large number of manufacturers [11,12]. Therefore, if these un- wanted variations in the voltage and current signal are not miti- gated properly [13], they can lead to failures or malfunctions of many sensitive loads connected to the same system, which may be very costly for the end users. To mitigate the PQ events, it is nec- essary to identify the events through PQ events detection and clas- sification system so that accordingly mitigation action can be carried out. Therefore, PQ events detection and classification area is having its own importance in the era of PQ. So, PQ analysis is becoming the most interesting area of research in past several years for characterization [14,15] and classification of events [16]. For the classification of PQ events, feature extraction and clas- sification are the most important part of the generalized PQ event classification system. PQ event detection requires the feature extraction from the input PQ disturbance. Feature extraction plays an important role in PQ analysis. Better feature set should be able to represent the signal efficiently. Extracted feature set can be used for the classification process. A brief survey on methods for PQ events classification reported earlier in [17,18] but little insight into the comprehensive analysis of different techniques. New and powerful tools for the analysis of PQ events diagnosis are currently available. Signal processing tools have been used for the feature extraction of power signal. Wavelets have been extensively used by the researchers during the last decade but other signal processing tools (i.e. such as S-transform (ST), Time-Time Transform (TTT), and Higher-Order Statistics (HOS)) have also been introduced to find out other salient features. 0142-0615/$ - see front matter Crown Copyright Ó 2012 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijepes.2012.04.045 ⇑ Corresponding author. Tel.: +91 130 2484124. E-mail address: itsmemanishkumar@gmail.com (M.K. Saini). Electrical Power and Energy Systems 43 (2012) 11–19 Contents lists available at SciVerse ScienceDirect Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes