Research On The Technology Of Gear Box Failure Diagnosis Based On Vibration Signal Analysis And Neural Network

Vibration signal analysis is widely used in the state monitoring and failure diagnosis of the gear and rolling bearing.This paper is for the purpose of studying several suitable methods for the gear and the rolling bearing vibration signal processing. Then use the neural network carries on pattern recognition after each kind of processing method processing data.First, this paper research the vibration motivation and vibration signal effect with different inspirit factor, research the signal characteristic of gears and rolling bearings vibrant. Then from time region, frequency region and time-frequency region introduce the gear and rolling bearing failure diagnosis method in detail. Finally, separately in the time region, the frequency region, determined several signal processing method carries on the simple diagnosis and the precise diagnosis.Secondly, with a foundation that understanding the vibration motivation and diagnosis method of gear bearing, built the gear box failure test installation, separately simulated the gear failure and the bearing failure to validate failure diagnosis method. Based on the limitation of each method in gear box failure diagnosis, especially the rolling bearing, This article propose a wavelet envelope demodulation analysis method, practice proof that compares with traditional method, the wavelet envelope demodulation analysis method is a better solution in the gear and rolling bearing failure diagnosis.Finally, carries on diagnosis using the BP neural network. The result shows this kind of neural network can withdraw the slight characteristic of input signal availability, can classification the defect with high accuracy and reliability.