Research On Key Technology Of Model-Free Predictive Control For PMSM Drive System

The dynamic and steady-state response of stator current and rotor speed is crucial to permanent magnet synchronous motor(PMSM)drive system.Therefore,the thesis conductes intensive research mainly on the model-free current control,the model-free deadbeat predictive control,the steady-state performance improvement of finite control set model predictive control(FCS-MPC)and the online optimization methodology of reference voltage vector of model predictive control,which has theoretical consequences and practical value for the control performance improvement of PMSM drive system in EVs.The current ultra-local model of surface-mounted PMSM(SMPMSM)drive system that consists of electrical parametric uncertainties,inverter nonlinearity and dq-axes coupling is first established through the stator current and reference voltage.Then,the model-free current control is proposed to improve the current control performance of SMPMSM drive system with the simultaneous disturbances caused by parametric uncertainties and inverter nonlinearity.After that,this thesis proposes model-free deadbeat predictive current control based on the discrete current ultra-local model which disengages from the dependence on mathematical model and improves in an overall manner the current control performance of SMPMSM drive system.Furthermore,the speed ultra-local model of SMPMSM drive system is established based on the rotor speed and reference torque,including mechanical parametric uncertainties,load torque,friction torque and torque tracking error,and the model-free deadbeat predictive speed control is presented.The dual-closed-loop control structure is composed by the combination of the model-free deadbeat predictive speed controlled outer loop and the model-free deadbeat predictive current controlled inner loop,and achieves the overall improvement of the speed control performance for SMPMSM drive system.In order to improve the steady-state control performance of conventional FCS-MPC,the optimal phase is obtained by establishing Lagrange interpolation polynomial between cost function and phase of reference voltage vector.Then,the amplitude of reference voltage vector is optimized based on the principle of minimizing the cost function,and the continuous voltage vector model-free predictive current control for SMPMSM drive system is proposal.Based on the previous research,the optimization of reference voltage vector is further transformed into the optimal solution of phase and amplitude.The cost function related to the phase and amplitude of reference voltage vector is derived,respectively,and the analytic solution of model-free predictive current control for SMPMSM drive system is obtained.Finally,the technology roadmap which consists of software simulation and motor bench testing is used in this thesis and the simulation and experiment are implemented to verify the novelty and practicality of the proposed methods.