Research On Diagnosis Method For Bearing Faults Of In-wheel Motor Based On WMM Information Fusion

In order to eliminate the hidden trouble during the operation,improve the performance of in-wheel motors and ensure the driving safety of the electric vehicle driven by in-wheel motors,bearing faults of in-wheel motor are taken as the breakthrough point,fault feature extraction and fault diagnosis as the main research content,a diagnosis method for bearing faults of in-wheel motor based on WMM information signal fusion is proposed,including:(1)Based on the bearing fault state of in-wheel motor,the real operation condition of in-wheel motor is simulated,and the corresponding bench test is designed to obtain the bearing fault state monitoring information(vibration signal and noise signal)and establish the fault diagnosis database.(2)By analyzing the state symptom information of bearing fault of in-wheel motor,a WMM based fusion method for monitoring information of bearing operation state of in-wheel motor is proposed.Firstly,the highly sensitive symptom parameters which can reflect the fault state of inwheel motor bearing are extracted by using the stable average discrimination rate(SADR),and then Weibull mixture model is used to fuse and expand the symptom parameters of vibration signal and noise signal and obtain sufficient highly sensitive symptom parameters fusion sequence as training data,which lays the foundation for the subsequent diagnosis model construction.(3)Based on Hidden Markov model(HMM)and dynamic Bayesian networks(DBNs),the fault diagnosis methods based on WMM-DBNs and WMM-HMM are proposed respectively to realize the dynamic diagnosis for bearing faults of in-wheel motor,which lays a theoretical foundation for online monitoring the operating state of in-wheel motor and ensuring the running safety of the drive system and whole vehicle.