Fault Diagnosis Of Permanent Magnet Synchronous Motor Based On Characteristic Signal Recognition

In recent years,permanent magnet synchronous motor has been widely used in power generation,aviation,new energy vehicles and other fields.Due to the bad working conditions of the motor in some special occasions,the load impact is large and so on,it is easy to occur various faults,resulting in huge economic losses.Based on the analysis of the faults on permanent magnet synchronous motor,it is a great significance to use artificial intelligence technology to diagnose the faults of permanent magnet synchronous motor and make appropriate maintenance measures in time to prevent accidents.Based on the finite element model of permanent magnet synchronous motor and stator current signal analysis,the demagnetization fault and short circuit fault of permanent magnet synchronous motor are analyzed.However the variational mode decomposition fault feature extraction of current signal,the permanent magnet synchronous motor is established based on probabilistic neural network fault diagnosis model,and the structure and parameters on the probabilistic neural network is optimized.The experimental results show that the proposed fault diagnosis for PMSM based on probabilistic neural network is effective.The main contents of this thesis are as follows:(1)By analyzing the structure of permanent magnet synchronous motor and the mechanism of common faults,the finite element model of PMSM is established,and the fault diagnosis method based on signal processing is analyzed,and the fault characteristic frequency of the current signal of PMSM is obtained.(2)In order to overcome the empirical mode decomposition in dealing with problems of mode mixing,the variational mode decomposition method is introduced.In view of the variational mode decomposition number selection problem in current signal decomposition,through spectrum contains similar frequency or the same frequency method to determine the number of decompositions.The energy entropy of the decomposed natural mode component is calculated,the validity of the method is verified by numerical simulation.(3)A fault diagnosis model of permanent magnet synchronous motor based on improved probabilistic neural network is proposed.For probabilistic neural network classification effect is influenced by structure complexity and parameters,the smoothing factor of the network is optimized by particle swarm optimization.And on the basis,the fuzzy c-means algorithm is used to get the clustering center to failure data,and the network model is rebuilt by selecting the samples closest to the clustering center as the neurons in the model layer of the probabilistic neural network.(4)The permanent magnet synchronous motor fault experiment platform is built,and fault diagnosis model based on improved probabilistic neural network is applied to the fault to verify the validity and accuracy of the model,and compare the model with the probabilistic neural network and probabilistic neural network for particle swarm optimization,the experimental results show that the fault diagnosis model of PMSM has higher accuracy.