Research On Disturbances Detection And Classification For Power Quality

With the rapid development of the long-distance, ultra high voltage and AC-DC interconnection in smart power grids, different types of power source and various forms electrical load access in the modern power system. Power quality is becoming worse and worse, which is causing the severe power quality problems. Power quality disturbance detection and classification is the key to improve and enhance power quality. In this paper, we analyzed and summarized the causes and harm of power quality problems, reviewed the research status of power quality disturbances classification, which combined with the current domestic and international power quality standards. In this paper, we research the method of disturbance detection and classification for power quality. The major works are as follows:1) In this paper, Ensemble Empirical Mode Decomposition and Hilbert transform were used for studying power quality disturbance signal. Firstly, aiming at the noise of power quality monitoring data, EEMD adaptive threshold denoising method was used. Compared with the 4 kinds of wavelet threshold denoising method, the simulation results show that the method is better denoising performance. And under the heavy noise pollution of low SNR, this method is also better performance. Then, EEMD method was used for a single or multiple power quality disturbance signals. PQD signals were decomposed into a number of independent intrinsic mode function IMF components. Hilbert transform can get the corresponding instantaneous frequency and amplitude. Thereby, the start and end time, disturbance frequency and amplitude information of PQD were detected. The simulation signal and measured data show that the method is effective.2) In this paper, Wavelet transform and S transform were used to extract the feature of power quality disturbances. Firstly, PQD signals were decomposed into different wavelet coefficients. We can get the normalized value decomposition by wavelet energy structures of different scales. It can make a preliminary classification by the feature wavelet energy spectrum. Then, S transform was used to analyze PQD signals. As the results of S transform for PQD, the amplitude feature of basic frequency, the variation feature of amplitude-frequency, the amplitude-frequency feature of standard deviation for high frequency, the amplitude fluctuations of basic frequency and the harmonic content were extracted. The five kinds of features can be used for subsequent identification of power quality disturbance recognition.3) In this paper, as for the single feature cannot express different PQD signals effectively, the combination of multiple features was proposed by analyzing the amount of complementarity between different features. According to the obvious difference of the wavelet energy for harmonic signal, we can identify it easily by setting a threshold. Then, combined with the 5 kinds of features of S transform, it can be used for subsequent identification of power quality disturbance. PSO-SVM was used to identify in this paper. The simulation results verify this method was effective and it provides a basis for the practical engineering application of power quality monitoring.