Automatic recognition system of underlying causes of power quality disturbances based on S-Transform and Extreme Learning Machine Hüseyin Eris ßti a,⇑ , Özal Yıldırım b , Belkıs Eris ßti a , Yakup Demir c a Electrical and Electronics Engineering Department, Engineering Faculty, Tunceli University, Tunceli, Turkey b Computer Engineering Department, Engineering Faculty, Tunceli University, Tunceli, Turkey c Electrical and Electronics Engineering Department, Engineering Faculty, Firat University, Elazig, Turkey article info Article history: Received 5 June 2013 Received in revised form 2 April 2014 Accepted 6 April 2014 Keywords: Power quality events S-Transform Extreme Learning Machine Classification abstract In this paper, a new S-Transform and Extreme Learning Machine (ST–ELM)-based event recognition approach for the purpose of classifying power quality (PQ) event signals automatically has been pro- posed. In this approach, the distinctive features of the PQ event signals have been obtained with the S-Transform-based feature extraction. The feature vector obtained with feature extraction has been applied as input to the ELM classifier. Ten different classification procedures were determined within the framework of this study to assess the performance of the ELM classifier on PQ event data. Real PQ event data and synthetic PQ event data obtained from MATLAB/Simulink environment have been used in these procedures. Also, three different PQ event data sets, which are formed by adding noises of 20, 30 and 50 dB to the synthetic PQ event data respectively, have been used in order to assess the perfor- mance of the proposed approach on noisy conditions. According to the results of performance evalua- tions, the proposed ST–ELM-based PQ event recognition system has a very high performance of recognizing PQ event data. Besides, classification of noisy data showed that the proposed approach is robust at recognizing noisy data. The performance of the ST–ELM-based recognition system on PQ data shows that this approach has an effective recognition structure that can be used in real power systems. Ó 2014 Elsevier Ltd. All rights reserved. Introduction Rapid developments in the energy industry have brought the studies on the quality of the power consumed to the fore. Power quality (PQ) has gained significance for energy providers and con- sumers recently. Undesirable situations that arise as a result of PQ problems affect much electrical hardware negatively. Also, poor power quality largely causes financial losses. According to the Euro- pean Power Quality survey report, PQ problems cause a financial loss of more than 150 billion Euros per year in the EU-25 countries [1]. The PQ-related problems generally happen due to the wide- spread use of non-linear and power electronically switched loads, lighting controls, unbalanced power systems, solid state switching devices, computer, industrial plant rectifiers and inverters [2]. Such PQ disturbances as instantaneous interruptions, sag, swell, harmonics and transients, resulting in switching, fault, lightning strike, etc. lead to a poor PQ. Identifying and classifying PQ problems play a significant role in reducing these problems. The methods in the literature to classify PQ problems generally fall into two groups. The first group is called the classification of PQ disturbances. The disturbances in this group are classified typi- cally as voltage sags, voltage swells and interruption [2–10]. The sec- ond group among the suggested methods in the literature is event classification. The causes underlying the disturbances such as faults, capacitor switching and transformer energizing are classified in this group [11–16]. The methods used in recognizing and classifying power system problems basically involve the stages of pre-process- ing, feature extraction and classification. The feature extraction of signals can be performed by direct techniques, such as the RMS value [16] of the raw samples, or transformation techniques, such as the Fourier transform [9], the wavelet transform (WT) [4,5,10,13,15], Kalman filter [7,14], Hilbert Huang Transform [12], correlation fea- ture selection [11] and S-Transform (ST) [2,3,6,8,17–19]. One of the advantages of using ST instead of wavelet for feature extraction is that it removes the necessity of testing various wavelet families in order to select a suitable wavelet family for the best classification [20]. The superior properties of S-Transform can provide significant improvement in the detection of PQ disturbances [17]. In this paper, the ST method, which presents considerable advantages in feature extraction of PQ event signals, has been preferred. When classifying http://dx.doi.org/10.1016/j.ijepes.2014.04.010 0142-0615/Ó 2014 Elsevier Ltd. All rights reserved. ⇑ Corresponding author. Tel.: +90 428 2131794; fax: +90 428 2131861. E-mail addresses: huseyineristi@gmail.com, heristi@tunceli.edu.tr (H. Eris ßti). Electrical Power and Energy Systems 61 (2014) 553–562 Contents lists available at ScienceDirect Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes