Research On Classification Of Power Quality Disturbances

Power quality is one of the research focuses in power system. With the widely used of power electronics, power pollution becomes more and more serious, while various utilization equipments based on the computers and microprocessors require higher power supply quality. In order to statistically analyze the power quality problems caused by power quality disturbances, power quality disturbance signals must be monitored and recorded, and it's important to the evaluation of the voltage quality that disturbance signals can be classified automatically by extracting the features from those recorded data.In this thesis, first the related concepts of power quality and it's related analysis methods is introduced, then a time-frequency analysis method, S-transform is used in the power quality analysis. S-transform succeeds to the time-frequency character of wavelet transform, and it transforms the time series into a time-frequency matrix, which contains the magnitudes, and phase information of the signals and provides evidence for the classification of power quality disturbances.At first the types of concerned disturbances include harmonics, flickers, voltage sags, voltage swells, voltage interruptions and impulsive transients. The S transform is used to generate the module time-frequency matrixes. Considering the data of the fundamental frequency and the high frequency, five indices which are the average value of the fundamental frequency amplitude, the maximum value of the fundamental frequency amplitude, the variation degree of the fundamental frequency amplitude, the minimum value of the signal and the average value of the high frequency amplitude are presented. Analysis results showed that by these five indices, features of different disturbances can be extracted effectively. The proposed approach provides a new idea for power quality analysis. Then the types of disturbances discussed include a combination of short-term disturbances and long-term disturbances, as well as their homologous single ones. Concerning the five indices discussed above, two more indices which are the symmetry degree of the fundamental frequency amplitude and the maximum value of the high frequency amplitude are presented. A simple decision tree is built to classify power quality disturbances. Power quality disturbance signals used for testing are all kinds of simulation examples containing random parameters, and their mathematical models meet the IEEE standard, so that the effectiveness of this method can be tested comprehensively. The process of the decision tree is clear. The indices can be used to classify different kinds of disturbances clearly, which can be calculated simply so that the computation time of the classification process can be reduced effectively. Numerical results show that the method proposed can effectively classify different disturbance patterns and provide basis for power quality mitigation measures.