Analysis of power quality disturbance based on improved HHT and BPSO-SVM

Aiming at the problems of complex feature extraction and low classification accuracy of traditional composite Power quality disturbance (PQD) classification methods. This paper combines I. Hilbert-Huang transform (HHT) and Binary Particle Swarm Optimization Support Vector Machine (BPSO-SVM) to analyze PQD. The different disturbance components in the original signal are decomposed into different Intrinsic Mode Function (IMF) components by Empirical Mode Decomposition (EMD) decomposition, to transform the multi classification problem into a binary classification problem to improve the classification efficiency. Aiming at the problem of modal mixing in the process of EMD decomposition, the frequency information in the same octave that is far apart in the same octave is stripped by adding a preset correction signal to the signal in the decomposition process. The simulation results of Matlab show that the accuracy of the compound disturbance classification is significantly improved.