Applied Study of Electromotor Fault Diagnosis Based on Wavelet Packets and Neural Network

Abstract

This paper presents a novelty electromotor fault diagnosis method based on wavelet packets and BP neural network. Wavelet packets transform can decompose original signal into different spectrums, the corresponding energy eigenvector can be obtained, it expresses the energy characteristics of original signal; BP (back propagation) neural network is an effective tool to recognize fault types, neural network can be trained by sample signal, then it can recognize practical fault types. The practical example indicates the validity of the new method.