Research On Tolerant Sequential Model Predictive Control And Adaptive Error Correction Algorithm Of Permanent Magnet Synchronous Motor

Permanent magnet synchronous motor(PMSM)has advantages of reliable operation,high power density and high efficiency,and has been widely used in electrified transportation and other fields,especially in new energy vehicle and traction train.However,the multivariable,nonlinear,and strong coupling characteristics of PMSM make it difficult for traditional control strategies to flexibly fit many applications.In order to meet requirements of complex working conditions,system often needs to precisely control multiple different objectives under nonlinear constraints to improve the comprehensive performance.Among various common intelligent control strategies,model predictive control(MPC)is based on modern control theory,with simple and intuitive principle,can control multi-objectives,but its weighting factor debugging process is cumbersome.In addition,in the PMSM closedloop control,the feedback of state variable phase current is very important,and in the nonideal situation,current measurement errors(CMEs)will cause torque and speed to pulsate periodically,reducing control accuracy.Focusing on the torque control of PMSM,this paper studies weightless model predictive control,and is committed to correcting current measurement error that causes torque ripple.The specific works are summarized as follows:Firstly,starting from the establishment of the PMSM mathematical model,the control variable of the motor system,that is,the inverter voltage vector’s action mechanism is analyzed,and the essence is to control the motor electromagnetic torque.In addition to the ideas and principles of two traditional motor control strategies,vector control and direct torque control,common motor intelligent control strategies such as MPC are also introduced to lay the theoretical foundation for subsequent research.Secondly,in order to solve the weighting factor problem in PMSM-MPC multiobjective optimal control,the predictive model of PMSM is expounded,the definition of multi-objective optimization problem and its Pareto optimal solution are introduced.Based on two traditional solutions,weighted scalar method and lexicographic method,the implementation processes of traditional MPC and sequential MPC are described by taking torque control as an example.Combined with the weak lexicographic method,tolerant sequential MPC(TSMPC)is proposed,and its feasibility and stability are analyzed.Build Matlab/Simulink simulation model to compare the performance of different MPC strategies.Then,for current measurement error issue in PMSM closed-loop control system,the dq-axis current formulas under offset and scaling errors are deduced respectively.Combined with deterministic mathematical features in the dq-axis current measurement error formulas,that is,both expressions are the weighted sum of sinusoidal signals,this paper adopts and improves an adaptive selection harmonic elimination(ASHE)algorithm and the dq-axis current compensation values are output online simultaneously.In the simulation model,the error correction effects of the adaptive algorithm before and after the improvement are verified.Finally,based on the PMSM control platform,the proposed TSMPC and adaptive error correction algorithm are experimentally verified.The results show that TSMPC can directly set the tolerance value instead of the tedious adjustment of weighting factors,and realize the multi-objective optimal control while achieving the best overall performance;the improved ASHE algorithm is simple to calculate,does not require motor parameters and additional sensors,can online correct current measurement errors and reduce speed ripple.Aiming at the high performance and high precision PMSM torque control system,this paper realizes the weightless MPC under multi-objective optimization and adaptive correction of current measurement errors,verifies the feasibility of the method.