Research On Fault Diagnosis Method Of Hub Motor Bearing Based On Sparse Representation

With the development of electric vehicles,the application of hub motor technology in electric vehicles is becoming more and more popular.The hub motor integrates the functions of driving,braking and bearing.It has the advantages of compact structure,flexible control and high transmission efficiency.It is a good choice for future electric vehicle technology.However,the driving conditions of electric vehicles are variable,and complicated road conditions will strongly impact the hub motor and induce mechanical fault.Hub motor bearings are important components for the rotation and load of the motor.They are both axially loaded and subjected to radial loads.It is the most vulnerable parts of the hub motor.Therefore,it is necessary to carry out the research on the diagnosis method of the hub motor fault while developing the hub motor technology.This paper starts from the simulation of the operating conditions of the electric motor hub motor under typical load.The real operating conditions of the hub motor and the simulation test under the fault condition Studying is studied.The vibration signal of the hub motor under normal and fault conditions are obtained.Based on the resonance sparse decomposition,the characteristics of the vibration signal under the fault state of the hub motor are analyzed,and the filtering method of the complex noise signal is studied to extract the fault characteristics of the hub motor.Based on the non-negative sparse representation,the characteristic parameter sensitivity of the fault characteristics of the hub motor is analyzed,and the sensitive feature parameter extraction method is studied to extract the sensitive characteristic parameters that characterize the fault state of the hub motor.Based on the random forest algorithm,the fault diagnosis model of the hub motor is constructed to identify the fault state of the hub motor.Firstly,the operating conditions of the electric motor hub motor under typical load are analyzed.Combined the test cost and existing equipment,the hub motor test bench,test control system and test signal acquisition system are built,the hub motor fault test plan is designed.Data for the study of hub motor fault diagnosis methods are provided.Secondly,it is difficult to select the decomposition parameters in the resonance sparse decomposition method.An RSK-based resonance sparse decomposition method is proposed.The method uses RSK as the optimization objective function,and uses the successive optimization algorithm to obtain the optimal decomposition parameters,and then uses the optimal decomposition parameters.The RSK-based resonance sparse decomposition method is compared with the traditional resonance sparse decomposition method.The simulation analysis and experimental results show that the RSK-based resonance sparse decomposition method can effectively filter out the noise components in the hub motor fault signal and extract fault features.Thirdly,the problem is selected for the characteristic parameters that characterize the fault state in the fault diagnosis.A sensitive feature parameter extraction method based on sparse representation of non-negative weight coefficients is proposed.Based on non-negative sparse principal component analysis,EM algorithm is used to solve non-negative sparseness.The principal component model optimizes the target with the inter-class dispersion as the parameter,and the sensitivity of the characteristic parameters is characterized by the non-negative weight coefficient,and the sensitive characteristic parameters of the different degrees of fault state of the hub motor are extracted.Finally,based on the random forest algorithm,the hub motor fault diagnosis model is constructed to identify the fault state of the hub motor.Combined with the hub motor fault test,the fault diagnosis method based on sparse representation is compared with other methods.The results show that the fault diagnosis method based on sparse representation can effectively extract the fault characteristics of the hub motor and accurately identify the fault state.