Research On Predictive Current Control Of Three-Vector Model Of Permanent Magnet Synchronous Motor Based On Parameter Identification

Permanent magnet synchronous motors(PMSMs)is widely used in wind power generation,national defense and other important fields because of their high power density and wide speed range.In recent years,model predictive control(MPC)techniques have been widely used in the field of PMSM control.MPC has a variety of advantages,such as simple control method,fast dynamic response,and easy cooperative control of multiple objectives.However,at the same time,it has the drawbacks of large computation and high parameter sensitivity.Based on the above key problems of MPC,the thesis takes model predictive current control as an example to investigate the reduction of its parameter sensitivity and computation,and the main research contents are as follows:Firstly,the PMSM mathematical model is constructed and the Eulerian dispersion method is applied to obtain the prediction model,and the traditional single-vector,two-vector,and three-vector MPCC algorithms are further investigated.The selection of voltage vectors in traditional MPCC is mostly done by enumeration method,which is computationally intensive.The first alternative vector is determined by judging the sector where the desired voltage vector is located,then substituting two voltage vectors within the sector into the value function,and finally combining the first alternative vector with the remaining five voltage vectors in turn and substituting them into the value function to determine the second alternative vector.The simulation results indicate that the optimized three-vector MPCC algorithm in this thesis not only reduces the computational effort,but also expands the coverage of alternative voltage vectors and improves the steady-state performance of the system.Secondly,parameters of the PMSM change with the operating conditions,and when the motor parameters are mismatched with the prediction model,it will lead to errors in the MPCC and thus degrade the system performance.According to the problem,an improved Adaline neural network algorithm is proposed in this thesis.The algorithm mainly improves the fixed-step LMS algorithm in order to solve the problems of large steady-state error and slow convergence under fixed step length.It is applied to motor parameter identification and combined with optimized three-vector MPCC.The simulation results indicate that the improved Adaline neural networkbased parameter identification algorithm in this thesis can effectively reduce the prediction errors of MPCC when the parameters are mismatched.Finally,the simulation and experimental results indicate that the PMSM threevector MPCC based on parameter identification in this thesis can improve the steadystate performance and computational efficiency of the three-vector MPCC,as well as eliminate the influence of parameter errors.Besides,reducing the parameter sensitivity of the three-vector MPCC.