Research On Fault Diagnosis Of Urban Rail Vehicle Bearings

In recent years,with the continuous increase of urban development and expansion in China,the problem of urban traffic congestion is becoming more and more serious,which becomes a huge problem of restricting urban development.The operation of urban railvehicles(subway and intercity railway)plays an important role in enhancing urban traffic capacity and alleviating road pressure.Timely and accurate fault diagnosis of urban rail vehicles can effectively reduce its possible serious consequences,saving cost and reducingpotential risk.According to relevant statistics,the mechanical failure mainly from the rolling bearing of urban rail vehicle,so the fault diagnosis of rolling bearings is essential.As one of the most widely used mechanical parts in urban rail vehicles,the signals of rolling bearingsare nonlinear and susceptible to noise.How to accurately extract fault characteristics and combine fault diagnosis method to analyze is important to ensure the safe and stable operation of vehicles.Aiming at the main fault types of rolling bearings of urban rail vehicles,the paper analyzes the characteristics of common fault signals of rolling bearings combining with traditional and intelligent fault diagnosis techniques at home and abroad.A fault diagnosis model of wavelet packet and adaptive genetic algorithm to optimize the least squares support vector machine(AGA-LSSVM)is proposed.what's more,the paper designs rolling bearing fault diagnosis system of urban rail vehicle with the help of virtual instrument technology and MATLAB,which realizing the diagnosis of common faults of rolling bearing,and the accuracy rate was higher.First,the vibration signal of the rolling bearing is subjected to wavelet denoising to reduce the interference of the noise.And the energy eigenvector of vibration signals which can reflect different fault types was extracted by using wavelet packet transform and time-varying autoregressive model(TVAR),then the extracted fault feature is inputed into the LSSVM model and BP network model which is optimized by adaptive genetic algorithm to identify the failure mode,and feedback the results to the monitoring interface to achieve intelligent fault diagnosis.The results shows that the system can accuratelyextract the waveform characteristics of the rail vehicle under different conditions,and the diagnostic accuracy is high,so it has certain engineering reference value.