A Method Based On Multi-Resolution Analysis Of Wavelet And Wavelet Packet Transform For Harmonics Analysis Of Power Quality

With the rapid development of national economics and electricity markets, both utility companies and their customers in many countries have paid close attention to power quality issues. A poor supply quality can degrade the quality of products, interrupt the important industry processes and therefore lead to economic losses.To improve the electric power quality via reasonable steps, a monitoring and analyzing system must be established to detect estimate and classify the different disturbance. Over the past two decades, much attention has been paid to deterministic harmonic analysis. However, there are a lot of non-stationary time-varying signals in electric power system, but some modern equipment are sensitive to this kinds of signals, so that they can't work well. Here, we study the detection methods in accordance with time-varying harmonics of power quality.Some Short-Duration Variations phenomena are illuminated in this paper. Studies of the harmonics and their theory foundation have been reviewed and discussed, and then the merits and demerits of these traditional methods are pointed out. According to the characteristics of time-varying signals, we introduce the wavelet transform fitting for detecting non-stationary signals, which breaks through the traditional methods. Wavelet multi-resolution analysis and wavelet package transform are applied to the analyses of many kinds of power qualities problems. Based on wavelet multi-resolution analysis and wavelet package transform, the method for harmonics analysis of power quality, is put forward in this paper. In addition, according to the characteristics of wavelet packet transform and that of conjugate quadrature filter bank, we bring forward the ameliorated arithmetic of wavelet packet. The amplitude of time-varying harmonics can be measured accurately through the ameliorated arithmetic. We also study the theory of wavelet transformand its application in detecting singularity and denoising, wavelet packet transform and its ameliorated arithmetic. After that a set of integrated method is shown as follow. Firstly, the saltation-point can be detected by using small scale wavelet transform; secondly, if there are serious noises on some occasion, the signals can be denoised by wavelet transform; thirdly, it can be judged whether the signals are the short-duration voltage variations (such as interruption, sag, swell) or the low frequency oscillatory transient by the wavelet multi-resolution analysis; lastly, the distinguished time-varying harmonics signals can be decomposed into their corresponding frequency band. All the processes are simulated by using MATLAB, and the results of the simulations are valid, which proves that our methods are reliable and practicable.