Research On Model Predictive Control Of Permanent Magnet Synchronous Motors

This paper is based on the Graduate Innovative Funding Program of Hebei Provice.On the basis of deep research on model predictive torque control(MPTC)and model predictive current control(MPCC)of permanent magnet synchronous motor(PMSM),this paper mainly focuses on the problem of weight factor design and computational burden of model predictive torque control(MPTC)for permanent magnet synchronous motor(PMSM),and novel strategies are proposed.The main research work are shown as follows.Firstly,aiming at solving the problem of weight factor design of MPTC,model predictive voltage control(MPVC)and model predictive flux control(MPFC)for permanent magnet synchronous motor(PMSM)are proposed respectively.For MPVC,it establishes a mathematical model of derivative of torque and stator flux,and the discretization is carried out.Through the equivalent transformation between the reference value,deduces the relationship between expected torque and stator flux amplitude and expected dq axis stator voltage,and takes dq axis the stator voltage as the control variable instead of the torque and stator flux amplitude.For MPFC,according to the mathematical equation of PMSM,the mathematical model of torque,flux and load angle is established.And then,by investigating the relationship among torque,stator flux and stator flux vector,the controlling of torque and stator flux amplitude is converted to the controlling of stator flux vector.Since the proposed MPVC and MPFC can automatically achieve the optimal balance between torque ripple and stator flux ripple,the weight factor in the cost function of traditional MPTC is eliminated,and better control performance is obtained.Secondly,for the traditional model predictive control(MPC),it is difficult to select the optimal voltage vector by enumeration method because of its large amount of calculation and limited practicability,and novel fast voltage vector selection,namely fast MPVC and MPFC,are proposed to solve the proplem.Firstly,the stator flux prediction model is deeply deduced,and the control of different variables is converted to the control of the voltage vector.By analyzing the process for voltage vector selection in detail,the proposed method only requires one prediction to select the optimal vector.The experimental results show that the proposed fast MPVC and MPFC can reduce the computational complexity and achieve better control performance.Finaly,the PMSM control system test platform based on 32 bit floating point TMS320F28335 digital processor is built,and the design of related hardware and software is completed.The experimental research and detailed analysis of DTC,MPTC,proposed MPVC and MPFC are carried out,and the comparison mainly includes: steady state performance,dynamic response performance,robustness and algorithm execution time.The results confirm that the proposed MPVC and MPFC can effectively suppress torque ripple and stator flux ripple while reducing the amount of computation,and it shows equivalent high control performance as MPTC in both steady state and dynamic state.