Research On Fault Feature Extraction Method Of In-wheel Motor Bearing Based On AHNs

In-wheel motor is one of the core technologies of "wheel drive" system of electric vehicle.It integrates functions of driving,carrying and braking,and has the advantages of flexible control and high transmission efficiency.The bearing applied to in-wheel motor not only has to bear the radial load but also the axial load,and is an important component of the in-wheel motor,which directly relates to the running performance of in-wheel motor.However,the complicated and variable driving conditions of vehicle and the unique installation position of in-wheel motor can easily induce mechanical failures of in-wheel motor bearing.For four-wheel independent drive electric vehicles,failures of single or multiple in-wheel motors will cause the output torque pulsate of in-wheel motor,which will cause the undesired angular acceleration of the vehicle and cause traffic accidents or even cause the life-threatening eventually.Therefore,it is urgent to establish a perfect in-wheel motor condition monitoring and fault diagnosis system.The in-wheel motor bearing is the vulnerable part of in-wheel motor.In this paper,in-wheel motor bearing was taken as the research object,and the fault characteristic signal extraction under intermittent strong interference was taken as the research content.A fault feature extraction method based on AHNs was proposed to effectively realize the fault feature extraction of in-wheel motor bearing under complex and variable operating conditions.Firstly,the operating environment of electric vehicle and the in-wheel motor structure were analyzed,the causes of the failure of in-wheel motor bearing were traced,and the main fault types of in-wheel motor bearing were summarized.Also,the fault characteristic frequency of the rolling bearing was expounded,which can provide a theoretical basis for the calculation of fault characteristic frequency of test bearing.Secondly,based on the consideration of the complexity of the actual driving condition of electric vehicle,the in-wheel motor fault simulation test bench was built,and the test plan that matches the actual running state of electric vehicle was designed to collect the vibration signal of in-wheel motor bearing under normal and fault conditions.Thirdly,characteristics of vibration signal of in-wheel motor bearing under complex driving conditions were analyzed.The fault feature extraction method based on AHNs was proposed and its implementation steps were established.Different forms of fault signals were established based on Matlab simulation software,and AHNs feature extraction method was used to extract the fault signals to verify its feasibility and effectiveness.The influence of AHNs structure parameters on AHNs feature extraction was studied based on simulation signals and the practicality of AHNs feature extraction method in engineering application was analyzed.Finally,the AHNs feature extraction method was used to extract the testing bearing vibration signal of in-wheel motor.The results showed that the method can separate the high and low frequency signals in vibration signal,filter out the interference signal,and extract the fault characteristics under intermittent strong interference eventually,so that realize the fault extraction of in-wheel motor.Also,the feature extraction results were quantitatively analyzed to further verify the effectiveness of extraction of bearing fault characteristics of in-wheel motor.The SNR value was used as the evaluation index to compare the method with the FIR method,the results indicated the advantages of AHNs fault feature extraction method.