Power Quality Disturbance Signal Denoising Based on Overcomplete Representation

Abstract

A denoising method for power quality (PQ) disturbance signal based on optimized decomposition of overcomplete dictionary is proposed. First, based on the idea of sparse decomposition, an overcomplete synthesis dictionary matching the time‐frequency characteristics of PQ disturbance signal is constructed. Then, the matching pursuit (MP) algorithm is used to search the disturbance parameters in Gabor dictionary, and the improved particle swarm optimization (PSO) algorithm is combined with the improved Nelder–Mead simplex extremum algorithm to improve the search speed and accuracy. Finally, the obtained disturbance parameters and the overcomplete synthesis dictionary are used to extract the disturbance characteristic waveform, and then the noise is separated. This method is used to reduce the noisy PQ signals and compared with the results of wavelet and other noise reduction methods, the results show that the method can effectively denoise PQ signal, and better retain the original characteristics of PQ disturbances. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.