Research on rolling bearing fault diagnosis of high-speed train

Abstract

The rolling bearing is the main vulnerable component of a rotating machinery on high-speed train, which is very important to the safe operation of high-speed train. Based on the fundamental analysis of the causes of rolling bearing faults on high-speed trains, the feasibility of wavelet packet analysis and BP neural network in rolling bearing fault diagnosis of high-speed trains is discussed. In this paper, wavelet packet is used to denoise data and extract features, then the training BP neural network is used to diagnose and analyze the characteristic data accurately, and the diagnosis results are obtained. By comparing the training results, the number of wavelet packet decomposition is determined, and the accuracy of BP neural network diagnosis is improved. In this paper, the analysis is carried out by using the bearing data of American electrical laboratory “Case Western Reserve University”, an efficient and fast method for rolling bearing diagnosis is presented, which can be applied to real-time fault diagnosis and monitoring of rolling bearing faults for high speed train.