Classification Of Power Quality Disturbance Based On Wavelet Transform And PSO-BP Neural Network

In recent years, power quality has become a problem which attracts more and more attention by the community. However, to enhance the power quality, it is the first thing to detect and classify the power quality disturbance signals and then take measures according to different ones.First this paper introduces the issues of power quality detailed and analyzes the existing classification methods of power quality disturbances.Then,this paper proposes an improved method based on wavelet transform (WT) and BP neural networks.Power quality disturbance signals are decomposed with wavelet multi-resolution analysis and the feature vectors are extracted through the coefficients at different levels. The feature vectors consist of two parts:the energy differences between Power quality signals and ideal voltage signal, the standard deviations of every level. This paper just extracts the feature vectors at particular levels.Then BP neural network is used for the classification of power quality signals and particle swarm optimization(PSO) algithm is used to improve BP neural network.This paper mainly focuses on nine kinds of disturbance signal,voltage swells,voltage sag,voltage interruption,voltage flicker,harmonic, harmonic and voltage swells, harmonic and voltage sag, harmonic and voltage interruption, harmonic and volage flicker.In this paper,MATLAB is used to construct the disturbance signals and classify the type.The simulation result shows that this methord has a good classification accuracy and good anti-noise ability.