Research On Fault Diagnosis Of Local Demagnetization Of PMSM Based On Wavelet Analysis

Permanent magnet synchronous motor(PMSM)has the advantages of high conversion efficiency,small size,compact structure,et al,and is widely used in the field of electric vehicles.However,due to various reasons such as aging,collision,chemical corrosion,et al,permanent magnets will have the risk of local demagnetization.Therefore,in order to ensure the operation safety of electric vehicles,it is of great significance to study the fault diagnosis of local demagnetization of PMSM.In the research of local demagnetization fault diagnosis,most of scholars study the degree of demagnetization.The local demagnetization fault diagnosis method of PMSM proposed in this topic can not only distinguish the degree of demagnetization,but also quickly locate the demagnetization position of permanent magnet.The main research contents of this topic are as follows:(1)Under local demagnetization failure,a faulty motor simulation model is established through Maxwell software.By comparing the phase voltage and the coil voltage signal,it is concluded that selecting the coil voltage signal as the fault diagnosis signal can not only realize the demagnetization degree diagnosis,but also realize the fault location.(2)By comparing the advantages and disadvantages of various wavelet denoising methods,the wavelet threshold denoising method is selected to filter the diagnostic signals collected in the experiment.(3)In the simulation analysis of local demagnetization fault location,the permanent magnet demagnetization position can be quickly located based on the relative position relationship between the encoder Z signal and the diagnostic signal.(4)In the simulation analysis of local demagnetization fault degree diagnosis,the diagnostic signal is decomposed and reconstructed by the method of wavelet packet.Then the energy value of the wavelet packet at multiple scales is extracted from the signal.Besides,the energy entropy of the wavelet packet is calculated according to the energy value,and then the least square algorithm is used for linear fitting.The simulation results show that the degree of demagnetization is highly correlated with the energy entropy of the wavelet packet,and the degree of demagnetization can be diagnosed according to the fitted detection model.(5)An experimental platform for fault diagnosis of local demagnetization of PMSM is constructed.The hardware and software design of the drive,control,data acquisition and host computer communication modules are carried out in the platform.On the experimental platform,no-load and load experiments are carried out under the five operating conditions of normal and partial loss of magnetism of 30%,50%,70%,and 100%.The experiments show the diagnostic method is reliable.The experimental results show that in five cases,the demagnetization position of the permanent magnet can be quickly located according to the relative position relationship between the coil voltage after wavelet threshold denoising and the encoder Z signal.The wavelet packet energy entropy is calculated using the coil voltage signal which is denoised by the method of wavelet threshold,and then the least square algorithm is used for linear fitting.At this time,the degree of demagnetization is highly correlated with the wavelet packet energy entropy,and the linear correlation coefficient is 0.97.The combined detection model can realize the diagnosis of the degree of demagnetization and achieve the expected effect.