Intelligent condition monitoring and fault diagnosis of a gearbox based on Artificial Neural Network

In this paper the vibration test system for the gearbox of mining machine, the wavelet denoising method, the artificial neural network's essential principles and its features, BP network structures model in the gearbox fault diagnosis are discussed. Tested vibration signals are disposed by the method of wavelet denoising and than as the inputs of BP neural network. By using classical BP neural network, four kinds of typical patterns of gearbox faults have been studied and diagnosed and satisfied results have been acquired. The research results indicate that BP neural network with the excellent abilities of parallel distributed processing, self-study, self-adaptation, self-organization,associative memory and its highly non-linear pattern recognition is an efficient and feasible tool to solve complicated state identification problems in the gearbox fault diagnosis simultaneously.