Study On Time-frequency Analysis Method Of The Power Quality Disturbance

With the development of modern economy and widely used of electronicequipments, power quality has attracted more and more attentions in the field of powerapartments and users. Because harmonic, frequency and voltage fluctuation haveseriously affected the normal operation of power system, the study on harmonic,voltage sag, voltage swell, and voltage interruption in the power system has been greatsignificance. Time-frequency analysis technique has been developed rapidly in recentyears. It has been widely applied into many fields. Especially, the excellentcharacteristics of the bilinear time-frequency analysis technology, which are energyintensity time-frequency concentration, frequency edge features, etc., have made greatcontributions on the analysis of non-stationary and nonlinear signals.In this paper, firstly the characteristic of power quality and several commonpower quality disturbance signals are described, and then several commonly usedtime-frequency analysis methods are summarized in signal processing. According tothe analysis of power quality disturbance signal, a novel method based on empiricalmode decomposition and Choi-Williams distribution is proposed. This method isapplied to simulate and analysis several common disturbance signal, such as thesteady-state power quality disturbance signal harmonic signals and transient powerquality disturbance signal voltage sags, voltage sags, interruption. This method can bealso used to extract the feature information of disturbance signal. Compared with othertime-frequency analysis methods, we find that the method can effectively suppresscross term of quadratic time-frequency distribution and retain characteristics of thequadratic time-frequency distribution.Finally, this paper summarizes traditional filtering method. Aimed at thedenoising problem in practical application, a novel denoising method based onempirical mode decomposition and Choi-Williams distribution is proposed. Weperform a Hilbert transformation on the intrinsic mode functions, which are obtained by empirical mode decomposition, to translate every intrinsic mode function intoanalytic signals which can be analyzed in frequency domain. According to thefrequency characteristics of noise signal, the high frequency components of intrinsicmode functions are filtered out, and the low frequency components is performed aChoi-Williams transformation to reconstruct energy distribution of original signal, andthus a complete time-frequency distribution of non-stationary signal is obtained. Thephysical significance of instantaneous frequency in the distribution is clearer, and itstime-frequency aggregation is better. This method is used in the power qualitydisturbance signal; it is proved having favorable denoising effects.