Identification of Flicker Source Using Continuous Wavelet Transform Abdolaziz Ashrafian (Khouali) and Hadi Moghadam Banayem and Aref Doroudi and Mehrdad Rostami Department of Electrical Engineering Shahed University Tehran, Iran {khouali& doroudi & rostami}@shahed.ac.ir Abstract— In this paper, a method based on continuous wavelet transform is suggested for calculation of flicker power. The flicker power can be utilized to identify the flicker direction to a flicker source with respect to a monitoring point. In our proposed method, using continuous Gaussian wavelet transform, pure flicker waveforms are extracted from the measured voltage and current signals. The flicker power can then be calculated. The direction to a flicker source is obtained from the sign of this flicker power. The proposed algorithm is tested using IEEE 13 Node Test Feeder. The simulation results validate the effectiveness of the proposed method. Keywords- Power quality; Flicker direction; Flicker power; Continuous wavelet transform. I. INTRODUCTION Voltage flicker is created due to fast load variations in power systems. Motors, electric welders, rolling mills, electric arc furnaces and wind turbines are the main resources of the flicker [1]-[3]. Random variations of these loads results in a waveform with irregular envelop. The frequency of the flicker waveform is lower than system frequency (between 0.5 to 35 Hz) and its amplitude fluctuates between 90 to 110 percent of nominal voltage [4]. These voltage changes may result in light fluctuations and noise in television broadcasting and some effects on ICU and CCU systems [5]. The voltage flicker may exist in any power grid. The measured flicker level at the point of common coupling (PCC) is sum of the flicker level from the different sources. In costly mitigation processes, it is essential to trace the dominant flicker source. Many techniques have been applied to determine flicker contributions at the PCC. In [6], the low frequency variations of the PCC voltage and the load current are represented by the complex phasors. The phasor of the load current are then decomposed into a conforming current phasor which is in phase with the voltage phasor and a non-conforming current phasor represents the rest of the load current. If the network is assumed linear, the flickers related to the conforming and non-conforming currents are originated from the network side and the load side, respectively. One possible disadvantage with this method is that it requires a phase sensitive measurement, which is quite difficult to obtain with high accuracy. Flicker direction in [7] and [8] is determined by calculating flicker power and its sign. There are some deficiencies in this method which can lead to improper results, e.g. the method uses square [7] and envelope detectors [8] for demodulation process. These demodulation methods are non-linear operations and produce unwanted signal components within the flicker frequency range and consequently the flicker power cannot be calculated accurately. In this paper, a new method is proposed to solve the above problems. The method uses the wavelet transform which is an appropriate tool to analyze non-stationary signals and power quality disturbance analysis [9], [10]. Gaussian Continuous Wavelet Transform (CWT) will be used to determine amplitude and frequency of voltage and current flicker waveforms. These values will then be utilized to calculate flicker power and to identify the flicker source. Some useful features of the proposed method are listed below:  It does not produce unwanted signal components within the flicker frequency rang.  The computational burden is reduced.  The method can deal both frequency and time domain information simultaneously and with proper resolution.  If the measured signal contains several flicker waves with different frequencies, amplitude and frequency of each wave can easily determine separately.  Besides determining flicker direction, knowing the flicker sensitivity coefficients, the method owns the potential of being extended to calculate severity of the flicker ( 10 V  ) and compare it to the standard values. The paper is organized as follows: section II outlines the wavelet transform, section III presents wavelet demodulation procedure, section IV demonstrates the flicker direction determination approach, section V shows the simulation results and section VI draws the conclusions. II. WAVELET TRANSFORM With the advent of wavelet transform, this technology is being comprehensively applied in electrical power engineering. This emerging technology does not have the drawbacks of Fast Fourier Transform (FFT) where a window is used uniformly for spreaded frequencies. The wavelet transform presents a multi-resolution analysis and uses short windows at high frequencies and long windows at low frequencies to more closely monitor the characteristic of non-stationary signals such as voltage flicker generated signals. Given ) (t x as a time variable signal, the continuous wavelet transform is derived as (1) [11]: dt a t t x a a CWT ) ( ) ( ) , ( 2 1            (1)