A Method to Identify PQD Based on SVM and Wavelet Energy Distribution

Approached a method to identify power quality disturbance (PQD) type based on support vector machine(SVM) and improved wavelet energy distribution. Firstly, using wavelet transform to analyze PQD signals, extracting disturbance lasting time and energy differences of each level between PQD signal and standard signal as feature vectors, forming the training samples and testing samples. Secondly, pre-process the training set by using neighbourhood rough set model to delete those abnormal samples and disturbances. Lastly, train the PQD samples by using binary tree SVM (BT-SVM) to identify PQD signals. Simulation results indicate that the proposed method can identify seven PQD signals and sinusoidal signal, having an excellent performance on correct ratio(the average ratio can reach 92.03 percent), having high identify speed and strong resistance to noise, and is very suitable for PQD identification system.