Local Demagnetization Fault Recognition Research Of Permanent Magnet Synchronous Linear Motor Based On PSO-LSSVM

Permanent Magnet Synchronous Linear Motor(PMSLM)has the advantages of quick response,high positioning accuracy and low operating noise due to its simple structure and direct drive motion.Therefore,PMSLM is widely used in high precision machining fields such as laser cutting machine,CNC machine tool with less cutting force and glass substrate testing equipment.The consistency of permanent magnet magnetic field is the key factor for the stable operation of PMSLM.It directly determines the quality of products processed by high precision machine tools.However,most permanent magnets in PMSLM are made of NdFeB.Under the influence of external magnetic field interference,excessive peripheral temperature,chemical corrosion,long-term natural aging and other factors,permanent magnets are prone to irreversible demagnetization.When one or more permanent magnets in PMSLM secondary are demagnetized,the consistency of the permanent magnet magnetic field is destroyed,local demagnetization fault occurs in the motor,and the thrust fluctuation of the motor is intensified.Therefore,this subject aims to accurately identify the location and degree of demagnetization of permanent magnet in local demagnetization fault of PMSLM,and to realize the diagnosis of local demagnetization fault of PMSLM by combining the particle swarm optimization least square support vector machine(PSO-LSSVM)classification and recognition algorithm.This topic is mainly studied from the following four aspects:1.Based on the induced electromotive force(EMF),an analytical model of PMSLM local demagnetization failure is established.According to the topological structure of the double-sided PMSLM studied in this paper,the coil,permanent magnet and other components of the motor are introduced in detail.By using the method of equivalent strength magnetization,the analytical expressions of induced EMF under normal and local demagnetization state of the motor are derived.Thus,the analytical model of PMSLM local demagnetization fault based on the induced EMF is established.2.A multimodal analytical model including fault feature information is established,and parameters such as peak number,peak starting position,peak ratio,and demagnetization degree are extracted to construct the characteristic vector.By taking the difference between the induction EMF of the normal and the local demagnetization of the motor and making the absolute value processing,a multimodal graphs analytical model is established to extract the characteristic parameters of the local demagnetization fault of the motor.Considering the relative operation relationship between the permanent magnet and the coil in the motor and the characteristics of the corresponding multi-peak graph under different demagnetization states,this subject selects 7 types of 15 kinds of faults for the classification and identification of local demagnetization faults of the motor according to the relative position relationship of the four pairs of permanent magnets.The fault presetting is carried out from two aspects of demagnetization position and demagnetization degree of permanent magnet so as to obtain multimodal graphs corresponding to different fault types.By further comparing the multi-peak graphs,the characteristic parameters representing various fault types were extracted,and the corresponding relationship between the characteristic parameters and demagnetization types was obtained.3.Aiming at the 7 types and 15 kinds of fault types designed in this subject,PSO-LSSVM classification algorithm is int:roduced to realize the accurate identification of local demagnetization faults of PMSLM.After taking demagnetization degree into consideration,each fault type corresponds to multiple combinations of demagnetization degree.Therefore,combined with the characteristics of PMSLM local demagnetization fault types and the differences between the characteristic quantities,and based on the idea of tree structure,7 sub-classification models are established according to the corresponding relations between the characteristic factors and fault types.The complexity of classification algorithm is reduced and the accuracy of classification recognition is guaranteed.4.The finite element simulation model and prototype experimental platform are established.According to the 15 fault types in this subject,fault preset is carried out,and the characteristic values of simulation and prototype experiment are extracted respectively for verification.The results show that the accuracy of PMSLM local demagnetization fault identification is 100%.Therefore,the research on the local demagnetization fault identification of PMSLM based on PSO-LSSVM carried out in this subject can accurately identify the location and degree of demagnetization of permanent magnets.This paper presents a new reference method for off-line testing the consistency of secondary permanent magnet magnetic field in the process of motor manufacturing and periodic maintenance.