The Research On The Detection Of Transient Power Quality Disturbance

With the development of national economy and the improvement of people's living standards, transient power quality problems have become the common concerned issues of power sectors and users. The traditional methods are not suitable for handling non-stationary and transient disturbance signals of power quality. Therefore, developing a new analysis method that can detect and identify the transient power quality disturbances rapidly and accurately, has important significance. In this paper, order morphology and fractal theory are used to de-noise and detect the transient power quality disturbances.Transient disturbance signal often contains a lot of noise, which have a great impact on the detection, so it is necessary to de-noise the transient disturbance signal. A variety of methods of de-noising transient power quality disturbance signal are analyzed, and then this paper presents an adaptive order morphological filter for pre-processing transient disturbance signal. The adaptive process of percentile is implemented based on a minimizing criterion of the mean absolute error. It can achieve a good de-noising effect. This method not only can filter out both gaussian white noise and impulse noise at the same time, but also can maintain the singular points.Commonly-used detection methods of the transient disturbance are compared in this paper, as well as their respective scope of application and shortcomings are pointed out. Grid fractal is presented to detect the transient disturbance signal after the de-noising. According to the characteristics of the transient disturbance signal, combined with the definition of short-term grid fractal dimension, transient disturbance can be located fast and accurately through the analysis of the account of grid. The detection method is simple. It overcomes the shortcomings of the previous methods, as well as can detect the transient power quality disturbances in real-timeSimulation results on MATALB show that both filtering algorithm and detection method are correct and effective.