Torque Ripple Learning Diagnosis And Fault-tolerant Control Strategy For Demagnetization Fault Of Permanent Magnet Synchronous Motor

As the core component of AC speed regulation in electric drive systems,permanent magnet synchronous motors are widely used in military,aerospace,and industrial fields because of their low power consumption,high power density,and good control ability.Compared with traditional excitation motors,permanent magnets,the key components of permanent magnet synchronous motors,are susceptible to humidity,high temperature,chemical corrosion and other complex and changeable operating environments,and there is an irreversible risk of demagnetization,resulting in serious degradation of motor performance,which greatly limits The promotion and application of permanent magnet synchronous motors.In order to ensure the efficient and reliable operation of the permanent magnet synchronous motor and its drive system,the research on the fault diagnosis and fault tolerance control of the permanent magnet synchronous motor is particularly important.For the demagnetized fault diagnosis of permanent magnet synchronous motors,the core lies in how to extract key characteristic signals from the running motor,and to achieve accurate fault identification in a complex and changeable non-stationary state.This paper proposes an extreme learning machine diagnosis method based on fuzzy theory for the motor torque ripple,which realizes the online diagnosis of the uniform loss of magnetism fault of the permanent magnet synchronous motor under vector control.In addition,because the PID adjustment method in the traditional vector control is difficult to meet the effective fault tolerance under the demagnetized fault condition in the high-performance servo system,based on the double closed-loop control of the speed and current,the prediction control of the deadbeat current based on the flux linkage observation is studied.The main research work of this paper is as follows:（1）The main types and basic structure of permanent magnet synchronous motors are briefly discussed,and the mathematical models and vector control principles of the motors under different coordinates are explained.Secondly,based on the time-stepping finite element analysis method,the field-circuit coupling simulation design of the permanent magnet synchronous motor vector control system is studied.With the help of ANSYS Maxwell and Simplorer industrial simulation tools,a permanent magnet synchronous motor finite element field-circuit coupling joint simulation model is constructed.Maxwell is used to generate the motor demagnetized model and analyze the transient electromagnetic field of the motor under the demagnetized condition.Simplorer is used to construct the vector control algorithm,and realize the self-starting,variable speed and load of the surface-mounted permanent magnet synchronous motor through the cooperation of the two technologies.Run simulation response to obtain characteristic information such as stator current,back electromotive force,air gap flux density and torque ripple in different field loss faults under the double closed-loop control system of the motor.（2）Aiming at the fault diagnosis of loss of excitation of permanent magnet synchronous motors,a fuzzy limit learning machine diagnosis method based on torque ripple is proposed.First,the wavelet threshold denoising method is used to denoise the acquired torque data,and the wavelet packet transform is used to decompose the torque feature in three layers.The feature vector is constructed around the energy spectrum of different frequency bands,and the feature vector is input to the extreme learning.The machine conducts training and learning to realize the real-time diagnosis of the loss of magnetism fault of the permanent magnet synchronous motor.In view of the failure of extreme learning machine in the face of unbalanced data,the diagnostic model introduces fuzzy theory into the extreme learning machine for optimization training.After comparing the diagnosis results with traditional classification algorithms,it verifies that the permanent magnet synchronous motor based on the fuzzy theory extreme learning machine fails.The accuracy and validity of the magnetic diagnostic model.（3）Aiming at the problem of the loss of magnetism of the permanent magnet synchronous motor that causes the load capacity of the motor control system to decrease and the control failure,a predictive control scheme based on sliding model Lomberg observer cooperative deadbeat current is proposed.First,use the principle of flux linkage estimation of the sliding film-Lomberg observer to establish the real-time flux linkage observation formula of the permanent magnet of the permanent magnet synchronous motor,and then obtain the state current observation value through the analysis of the flux linkage formula,and finally input the state current observation value into In the pre-built deadbeat current fault-tolerant controller,the purpose of fault-tolerant control of the loss of magnetism fault of the permanent magnet synchronous motor is realized.The simulation experiment of deadbeat current prediction and fault tolerance based on MATLAB/Simulink verifies the feasibility and stability of the control method under the condition of demagnetization,and improves the control performance of the drive system.