Partial Demagnetization Fault Diagnosis Research Of Permanent Magnet Synchronous Motor Based On PNN Algorithm

Permanent Magnet Synchronous Linear Motor(PMSLM)is widely used in the field of laser cutting machine and few cutting force CNC machine tools because of its simple structure,high positioning accuracy and reliability,low noise and so on.As an important part of the PMSLM,the permanent magnet has an important effect on the output performance of the linear motor.However,the magnetic characteristics of PMSLM permanent magnet are easily affected to irreversible demagnetization of permanent magnets due to environmental conditions such as load conditions,temperature and aging of the motor,resulting in partial or uniform demagnetization failure of PMSLM.In particular,local demagnetization will produce an uncompensated deterioration in the output performance of the motor.Therefore,aiming at the local demagnetization problem of PMSLM,this paper proposes a local demagnetization fault classification and identification method based on the combination of spatial air gap flux density reconstruction and PNN algorithm,to achieve precise and rapid diagnosis of permanent magnet demagnetization fault location,it can ensure the safe,high-precision and reliable operation of PMSLM and also laid the theoretical foundation for the follow-up motor condition monitoring and fault-tolerant compensation of demagnetization fault.Aiming at the problem of partial demagnetization fault detection and classification in PMSLM permanent magnet,this paper carries out the following research.(1)According to the structural characteristics of PMSLM and the influence of permanent magnet magnetic field on its performance,the equivalent magnetization method was used to establish analytical model of space air gap flux density and induced electromotive force under normal and fault conditions.The correlation between the air-gap flux density the no-load back-EMF and the demagnetization of the permanent magnets are compared and analyzed.It is revealed that that space air gap flux density is sensitive to demagnetization of permanent magnets.It is theoretically proved that it is feasible to select space air gap flux density as information acquisition carrier for demagnetization fault detection of permanent magnets.The finite element model of PMSLM was established and the correctness of the analytical results was verified.The method of three parallel gas-density lines was proposed to further analyze the local demagnetization faults by finite element simulation analysis.(2)The finite element method is used to calculate 15 demagnetization types of three locations,which include air-gap centerline,above air-gap centerline and below air-gap centerline.The air gap flux density on the three lines is preprocessed and reconstructed as the unique feature that can identify the type of demagnetization fault.In order to eliminate the effect of the demagnetization degree of the permanent magnets on the classification results at different times,the fault characteristics of the permanent magnets with different degrees of demagnetization were normalized.And a fixed sample library of demagnetization fault was set up based on the number of fault features and the number of cycles.It laid the foundation for the identification of fault classification.(3)In order to identify the fault efficiently and accurately,This paper introduces the PNN multi-classification algorithm to overcome the insufficiency of BP.Compared with the classification results of BP network.The results show that in the equal training samples,PNN algorithm is faster and the classification recognition rate is higher.In order to get a better recognition effect,the PSO is introduced to further optimize the smoothing parameter 8 in the PNN to determine the classification result.A local demagnetization fault classification model based on the optimal parameters of the PNN network is established.Simulation and experimental results show that the optimized PNN network can achieve a recognition rate of 100%and can accurately diagnose the local demagnetization of permanent magnets.(4)A demagnetization prototype was designed according to the fault preconditions,a high precision Gaussian meter detection device was designed,and a PMSLM demagnetization fault density testing platform was set up.Four typical demagnetization faults were tested by a prototype and the measured fault Data classification results of the analysis.The simulation results and experimental results show that the proposed method can accurately identify the local demagnetization failure of PMSLM,and further verify the accuracy and efficiency of the proposed method.It not only enriched the demagnetization of linear motor diagnostic tools,but also for the same type of PMSLM demagnetization detection provides a new method.