Power Quality Disturbance Identification Method Based on Improved GSA-SVM Algorithm

Aiming at the power quality problems caused by the use of a large amount of power electronic equipment, nonlinear load and electrified railway in the distribution network, nine common models of power quality disturbance signals are built in MATLAB / simulink for simulation analysis. In this paper, a 10-layer fast wavelet decomposition method using db4 wavelet transform is proposed, and the energy values of detail components in each layer are calculated as eigenvectors. Aiming at the problem that the penalty factor and kernel function parameters of support vector machine (SVM) are easy to fall into local optimal solution in the course of optimization, an improved universal gravitation search algorithm (IGSA) is proposed to optimize the penalty factor and kernel function parameters of SVM. By optimizing the parameters, the IGSA-SVM classifier is constructed. The extracted feature vectors are normalized and input into the constructed IGSA-SVM classifier to train and identify the datas. The proposed method is tested by adding 0dB, 20dB and 30 dB Gaussian white noise to the signal, and compared with the GSA-SVM classifier. The results of simulation indicate that the proposed method is effective and precise.