1 Abstract--This paper presents a new methodology based on Independent Component Analysis (ICA) for power quality disturbance analysis. The proposed methodology aims at analyzing power quality disturbances that appear as mixtures in the voltage signal. Such disturbances are commonly referred to as multiple disturbances. Results are obtained from both simulated and experimental data showing that disturbance classification higher than 97 % can be achieved. The results are attractive for practical applications in power quality systems. Index Terms -- ICA, Power Quality, Multiple Disturbances. I. INTRODUCTION LECTRIC power quality has become an active research area in the last few years due to the growing concern of delivering clean power to consumers in the presence of distorted waveforms [1]. Waveform distortions are normally caused by disturbances such as voltage sag/swell with and without harmonics, momentary interruption, harmonic distortion, flicker, notch, spike and transients, causing problems such as malfunctions, instabilities, short lifetime, and failure of electrical equipments and so on [2]. According to [3], electric distribution network faults may cause voltage sag or momentary interruption whereas switching off large loads or energizing large capacitor banks may lead to voltage swell. On the other hand, the use of solid-state switching devices and nonlinear power electronically switched loads such as rectifiers or inverters may cause harmonic distortion and notching in the voltage and current signals. Flickers may be caused by the usage of arc furnaces and ferroresonance. Transformer energizing or capacitor switching may cause transients, whereas lightning strikes may lead to spikes. In this context, several signal processing and computational intelligence techniques have been proposed for power quality (PQ) monitoring [2]. These techniques try to achieve high performance with low computational complexity, which is required for online PQ monitoring. The PQ monitoring comprises, basically, two processing stages: disturbance detection and classification. Typically, this is developed using the voltage waveform. Recently, a D. D. Ferreira is with the Federal University of Lavras, UFLA, Lavras, MG, Brazil, emails: danton@ufla.br. Seixas is with the Federal University of Rio de Janeiro, COPPE/Poli, Rio de Janeiro, RJ, Brazil, e-mail: seixas@lps.ufrj.br. A. S. Cerqueira is with Department of Electrical Circuits, Federal University of Juiz de Fora, Juiz de Fora, MG, Brazil, e-mail: augusto.santiago@ufjf.edu.br bunch of methods have been proposed for the automatic recognition of the PQ disturbances [3,4,5]. These methods are capable of recognizing the PQ disturbances with promising accuracy rate. However, these methods aim at recognizing single PQ disturbance in a measured voltage waveform. Thus, the performance of these methods might be limited because, in real power systems, the disturbances could appear simultaneously. These events are commonly referred to as multiple disturbances. Some efforts have been done about this problem, where the works [6] and [7] stand out, but further studies are still required. This paper proposes a new methodology based on Independent Component Analysis (ICA) [8] for analyzing PQ events with multiple disturbances, following the idea first proposed in [9]. Here, ICA is applied in order to decouple the multiple disturbances envisaging the improvement of the classification efficiency. This paper is organized as follows. Section II formulates the multiple disturbance PQ problems. Section III presents the fundamentals of the ICA theory. Section IV describes the proposed method and Section V presents the achieved results. Conclusions are derived in Section VI. II. PROBLEM FORMULATION According to [6], monitoring the discrete version of powerline signals can be achieved through non-overlapped frames of N samples each. The discrete sequence in a frame is expressed as an additive contribution of several types of phenomena: () ()| : () () () () () s t nT vn vt fn hn in tn rn = = = + + + + , (1) where n=0,...,N-1, T s =1/f s is the sampling period, the sequences {f(n)}, {h(n)}, {i(n)}, {t(n)} and {r(n)} denote the fundamental component, harmonics, interharmonics, transient and background noise, respectively. Each of these signals is defined in details in [6]. So far, most classification techniques are developed for single disturbance analysis. Nevertheless, one can note that the incidence of multiple disturbances, at the same time interval, in electric signals, is an ordinary situation due to the presence of several sources of disturbances in the power systems. Figure 1 displays two typical cases. These voltage measurements were obtained from the IEEE working group P1159.3 website [10]. In Figure 1(a), one can note the incidence of a short-duration voltage variation named sag, ICA-based Method for Power Quality Disturbance Analysis Danton D. Ferreira, José M. de Seixas and Augusto S. Cerqueira E