Classification of Power Quality Disturbance Based on S-Transform and Convolution Neural Network

The accurate classification of power quality disturbance (PQD) signals is great significance for the establishment a real-time monitoring system modern grids, ensuring safe and stable operation electricity safety users. Traditional signal methods are susceptible to noise interference, feature selection, etc. In order further improve accuracy methods, this paper proposes method based on S-transform Convolutional Neural Network (CNN). Firstly, used extract obtain time-frequency matrix with characteristics signals. As an extension wavelet transform Fourier transform, can avoid disadvantages difficult window function selection fixed width. At same time, extracted by has better immunity. Secondly, CNN perform secondary extraction obtained high-dimensional modulus reduce data dimensions main features signal, then classified using SoftMax classifier. Finally, after series simulation experiments, results show that proposed algorithm accurately classify single different signal-to-noise ratios composite composed signals, it also good Compared other in timeliness higher accuracy, efficient feasible method.