The Research Of The Online Diagnosis Method For Bearing Fault Of In-wheel Motor Based On DBNs

With the development and application of in-wheel motor drive technology in passenger vehicles,four-wheel independent drive electric vehicles have more and more obvious advantages compared with traditional centralized drive electric vehicles,such as high transmission efficiency,low energy consumption and strong power.Therefore,in-wheel motor drive technology is considered as the future development direction of electric vehicles.In-wheel motor is the power source of electric vehicle,and its running state is directly related to the safety of the whole vehicle.Establishing a perfect running state monitoring and fault diagnosis system of in-wheel motor is one of the key technologies for the promotion of four-wheel independent drive electric vehicle market.Because of the special installation position and mode of in-wheel motor,the mechanical fault of in-wheel motor can easily be caused by the changeable driving conditions and complex road conditions.Once a mechanical fault occurs and the failure operation lasts for a long time,it will lead to the deterioration of the insulation material performance of in-wheel motor and abnormal friction between stator and rotor,which will lead to secondary faults such as winding damage,turn-to-turn short circuit and interphase short circuit.In serious cases,it will cause the loss of the performance of the whole in-wheel motor,affecting the handling and stability of the vehicle and endangering the safety of the driver and passengers.Therefore,on-line monitoring and fault diagnosis of in-wheel motor are urgently needed.In this paper,bearing fault,a common mechanical fault of in-wheel motor,is taken as the object of diagnosis.Based on vibration signal,highly sensitive parameters which can represent the fault state are extracted as the input of the later diagnosis model.Dynamic Bayesian network based on Gauss mixture output is used to build the mechanical fault diagnosis model of in-wheel motor to realize on-line diagnosis of mechanical fault.Firstly,the structure and working principle of in-wheel motor are expounded and the location and expression of common mechanical faults of in-wheel motor are analyzed.Combined with the installation position and the working condition of in-wheel motor in the electric vehicle,the three influencing factors that cause mechanical faults are summarized.Secondly,the selection of in-wheel motor is completed according to the performance index of electric vehicle called ZHIDOU D1.In addition to a normal motor,an in-wheel motor with outer ring fault is needed.Electric wheel clamp is designed and the in-wheel motor test bench is set up under the consideration of installation of in-wheel motor in the electric vehicle.The vibration signal acquisition scheme is designed to provide sufficient experimental data for training database and test database establishment of diagnosis model.Thirdly,the common parameters are calculated based on the vibration signals of in-wheel motor and Gaussian distribution test is carried out for calculated parameters;GDI value and SWDI value are proposed based on DI value.GDI is used to select the signal with more sensitive reflection of fault from the two measurement signals as the vibration signal selection method for diagnosis;Based on diagnostic vibration signal,the highly sensitive parameters are selected using SWDI value from the common parameters obeying Gaussian distribution.Finally,“speed slice” is defined to build the group of bearing fault diagnosis models for in-wheel motor based on Gaussian mixture output dynamic Bayesian network considering the operation condition of in-wheel motor.Then,an online diagnosis method for bearing fault is proposed based on the group of fault diagnosis models.The effectiveness of this method is verified by test database based on the highly sensitive parameters.The online diagnosis results show that the correct recognition rate is up to 98.3%.There is a good recognition of the normal state and the fault state of in-wheel motor and it can be used to provide theoretical basis and technical support for the safe operation evaluation of in-wheel motor.