Electric Power Systems Research 119 (2015) 425–431 Contents lists available at ScienceDirect Electric Power Systems Research j o ur na l ho mepage: www.elsevier.com/locate/epsr A method based on independent component analysis for single and multiple power quality disturbance classification Danton D. Ferreira a,∗ , José M. de Seixas b , Augusto S. Cerqueira c a Federal University of Lavras, Minas Gerais, MG, Brazil b Federal University of Rio de Janeiro, Signal Processing Lab, COPPE/Poli, Rio de Janeiro, RJ, Brazil c Department of Electrical Circuits, Federal University of Juiz de Fora, Campus Universitário, 36036 900 Juiz de Fora, MG, Brazil a r t i c l e i n f o Article history: Received 8 March 2014 Received in revised form 18 August 2014 Accepted 28 October 2014 Keywords: Electric power quality Independent component analysis Disturbance Classification Single channel Signal decomposition a b s t r a c t This paper proposes a method based on single channel independent component analysis for single and multiple power quality disturbance classification. The proposed method decouples the power system signal into its independent components, which are classified by specialized classifiers. The classifier outputs are combined by using a logic that gives the final classification. Five classes of single disturbances and twelve of multiple disturbances are considered and a classification efficiency above 97% is achieved for each event class. Both qualitative and quantitative analysis elucidate the efficiency of the proposed method. Results are obtained from both simulated and experimental signals. © 2014 Elsevier B.V. All rights reserved. 1. Introduction The quality of electric power has become an important issue for electric utilities and their customers. As a result, power quality study has become an active research area in the last few years [1]. Degradation in electric power quality is normally caused by dis- turbances such as voltage sag/swell with and without harmonics, harmonic distortion, notch, spike and transients, causing problems such as malfunctions, instabilities, short lifetime, failure of electri- cal equipments and so on. In order to determine the causes and sources of disturbances, one must have the ability to detect and classify these disturbances. These process in blocks are required for correct disturbance identi- fication before appropriate mitigating action can be taken. Recently, a bunch of methods have been proposed for the automatic recog- nition of the PQ disturbances, as reported in [1], however, most of them were proposed for recognizing single disturbances. Thus, the performance of these methods might be limited because, for many power networks, the disturbances may appear simultaneously and are commonly referred to as multiple disturbances. ∗ Corresponding author. Tel.: +55 3538291025. E-mail addresses: danton@deg.ufla.br (D.D. Ferreira), seixas@lps.ufrj.br (J.M.d. Seixas), augusto.santiago@ufjf.edu.br (A.S. Cerqueira). Regarding the classification of multiple disturbances, many recent works have considered the occurrence of only two classes: sag with harmonics and swell with harmonics [2–4]. The authors in [5] have used ant colony optimization technique for disturbance classification in which four classes of multiple disturbances were considered: sag or swell with harmonics, flicker with harmonics and interruption with harmonics. The authors in [6] have used a S- transform variant and Fuzzy decision tree for classifying six classes of multiple disturbances: sag or swell with transient, swell with harmonics, harmonics with notch or flicker and spike with tran- sient. In [7], sag or swell with harmonics, sag or swell with transient and sag or swell with flicker were combined allowing the classifica- tion of six classes of multiple disturbances by using fuzzy logic and particle swarm optimization. In [8], eight classes of multiple distur- bances (sag or swell with transient, sag or swell with harmonic, sag or swell with flicker, flicker with harmonic and harmonic with tran- sient) were considered for classification by using discrete wavelet transform and wavelet networks. These works have achieved good and promising results, however they are limited to a reduced num- ber of multiple disturbance classes (up eight) and combining only two types of single disturbances. In the present paper, a method based on single channel inde- pendent component analysis (SCICA) is proposed to analyze and classify PQ events with multiple disturbances, following the idea first proposed in [9]. In [9] the authors introduced the SCICA method to decompose and classify two classes of multiple disturbances: http://dx.doi.org/10.1016/j.epsr.2014.10.028 0378-7796/© 2014 Elsevier B.V. All rights reserved.