Research On Fault Diagnosis Method Of Motor Bearing Based On Parameter Adaptive VMD

As a widely used rotating machine,motor has become an indispensable electrical equipment in industrial production and people’s lives.Rolling bearing is an important part of motor.Once failure occurs during operation,it will affect the normal work of motor,causing serious economic losses.In severe cases,it will endanger personal safety.Therefore,it is of great theoretical significance and practical value to research on fault diagnosis method of motor bearing,which can find the faults in time and accurately,avoiding casualties and economic losses,and has important theoretical significance and application value.When the traditional signal analysis method is used to solve the problem of fault diagnosis,there is a lack of adaptive decomposition ability,relying on artificial setting parameters,and there are modal aliasing problems,which affect the fault diagnosis effect to some extent.Variational mode decomposition(VMD)is a new adaptive signal analysis method,which can adaptively realize the frequency domain division and the effective separation of each component of signals.It has a solid mathematical theoretical theory and good noise robustness.In this paper,the motor bearing is taken as the research object,aiming at the problem of the number of preset modes in VMD,multi-threshold center frequency method and kurtosis mean method are proposed to determine the number of VMD modes,combined with the signal power spectrum,and then a fault diagnosis method is proposed.The signal is decomposed by VMD to obtain the Intrinsic Mode Function(IMF).The fault characteristic frequency is extracted by solving the signal power spectrum of the IMF component,and the fault type is determined by the method of identifying the fault characteristic frequency.When the vibration signal contains a strong interference signal,in order to solve the problem that the fault characteristic frequency can not be extracted by the above method,the energy entropy and probabilistic neural network(PNN)are introduced.The energy entropy can accurately characterize the energy of a fault signal with frequency and the PNN has a small amount of calculation,fast convergence,and good classification ability for nonlinear problems.A motor bearing fault diagnosis method based on center frequency of multi-threshold method VMD energy entropy an PNN is proposed.First,the number of VMD modes is determined by the center frequency of multi-threshold method.Then the high-dimensional characteristic matrix is constructed by solving the energy entropy and finally input to the PNN.The fault type is determined according to the maximum output probability.The experimental platform was built and the proposed method was validated by the actual acquired vibration data.The experimental results show that the method can accurately identify the fault type accurately,and the effectiveness of the method is proved.