Detection And Identification Of Transient Power Quality Disturbances

In recent years, with the extensive use of power electronic loads, power quality has deteriorated., power quality issues highlighted. The power quality directly affects the economic development of social and the quality of life. Therefore, power quality problem has become a hot topic in the field of power systerns. Correct detecting classification, identification of power quality disturbances are very necessary.This dissertation discuss the automatic identification method of transient power disturbance based on dual-tree complex wavelet and support vector machine. This approach can use intelligent way to improve the recognition of transient power quality disturbances. Simulation results of this dissertation verified that the singular point of transient power disturbance signal can be positioned by the modulus maxima of wavelet coefficients. Meanwhile, the noise of transient power disturbance signal could be removed by the method of threshold to remove some high-frequency coefficients. Because of real wavelet transform extract limited information from transient power disturbance signal. But, the DT-CWT could provide useful information. So we choose the DT-CWT to analyse the signal. After DT-CWT, modulus maxima of the singular point is more obvious, and we can extract signal’s phase information. Wavelet coefficients of dual-tree complex wavelet can provide feature vectors to support vector machine. Feature vector can training support vector machine model and verify the accuracy of the classification model. In this dissertation, we select grid search optimization, genetic algorithms, PSO to search optimal parameters of support vector machine and select the classification parameters with highest accuracy.