The Detection And Recognition Method Of Power Quality Disturbances

With the demands for power quality increasing, the users are paying more and more attentions to the power quality disturbance. In order to improve and enhance the power quality, we must identify and detect the disturbances which effect power quality, and then we can find out the source of disturbance and take appropriate solutions. In this paper, two methods about power quality disturbance identification and detection are proposed: the one based on dq conversion and wavelet transform, the other one based on short-time fourier transform and singular value decomposition. The first method uses the features of dq conversion which can instantaneously get RMS voltage and wavelet coefficients which can characterize the abrupt information of disturbance. According to the differences of RMS, the size of mutation characteristic value and the spectral characteristics of some disturbances, we can achieve the goal of identifying and detecting the power quality disturbances. The second method uses short-time fourier transform to extract characteristic values from disturbances respectively, theses provide a quantitative foundation for the disturbance identification, and then classification tree is used to classify the disturbance. Meanwhile, this method proposes to transform one-dimension power quality signal to a matrix, then we can extract the mutation information from power quality signal through the singular value decomposition and achieve detecting the occurrence time of disturbance. Simulation results show that the proposed methods can implement classification of power quality disturbances and detection on the disturbance effectively and accurately, but the anti-noise performance of first method is worse, and can only analyze a single power quality disturbance. The anti-noise performance of second method is better, that is, it does not need the noise-suppressed processing on the general noise level, and also can analyze a variety of composite power quality disturbances. The noise is inevitable in the real grid, and disturbance is always occurring along with other disturbance. Therefore, in this paper, the identification and detection method of power quality disturbances based on short-time fourier transform and singular value decomposition has a strong practical value.