Research On Model Predictive Control Technology Of Permanent Magnet Synchronous Motor For Electric Vehicle

The permanent magnet synchronous motor(PMSM)has the advantages of high efficiency,high power density,etc.,and is most commonly used in the automotive industry as a traction motor for electric vehicles.Considering that the PMSM drive system is a multi-variable,nonlinear and strongly coupling system,it is extremely sensitive to parameters and disturbances.At the same time,the electric motor drive system has the characteristics of high voltage and large current,which are must be explicitly enforced.The emerging requirements for electric vehicle drive systems are to achieve fast dynamic response and provide high steady state control accuracy while meeting system constraints.Model predictive control has achieved great success in academia and industry,and has been a research hotspot.Its significant advantage is that it can systematically consider constraints when solving optimal control problems,allowing the control process to operate at the limits allowed.To this end,this dissertation focuses on the application of model predictive control theory to development the electric vehicle drive system.The corresponding model predictive control algorithms are designed for the current control and speed control of PMSM.The related research content content involves modeling and linearization,motor parameter identification,disturbance observer design,load torque estimation,and numerical optimization methods involved in implementing model predictive control.The design points of the model predictive control theory in the field of AC drive are summarized,and the parameter sensitivity in the current control problem of the permanent magnet synchronous motor is analyzed.In order to provide an accurate mathematical model for the MPC,combined with the characteristics of the prediction model,an online identification algorithm is introduced to identify the sensitive parameter.The identification results are used for back-EMF and cross-coupling terms calculation and target control voltage acquisition,and form the state parameter vector in the MPC scheme.Since the parameters which providing by the real-time identification algorithem that can reflect the time-varying characteristics of the plant.Therefore,the parameter adaptability has been effectively improved in the model predictive control algorithm strategy.The current controller is designed by using parameter identification combined with MPC theory.In addition,the explicit model predictive control technology is introduced to greatly simplify the calculation of the online optimization problem introduced by MPC.A robust current model predictive control is designed to cope with the parameter changes of the PMSM in actual operation.In the presence of constraints,the state and input constrained linear system model is obtained by augmenting the state disturbance term.An adaptive observer is designed to observe the disturbance term and state online.The combined online disturbance observer and model prediction control with inherent the zero-offset features in steady-state ensures the stability of the closed-loop system.The current controller is implemented with explicit model predictive control technology,which greatly reduces the online calculation effort.In order to overcome the difficulty of the traditional cascading structure of the speed controller and to achieve the optimal control objectives,a speed and current integrated model predictive control algorithm named LPV-MPC is proposed.Based on the analysis of the dynamic control model and the optimal current control strategy of PMSM,the state space model of PMSM dynamic is linearized at each steady-state operating point.The incremental control model of the rotor speed control is obtained by mathematical transformation.The extended kalman filter is used to estimate the unknown parameters such as load torque to form state feedback.The system incorporates the constraints into the iterative optimization process of MPC,and the system state is steered to the target setpoint through the penalty of the cost function.The control strategy has multiobjective optimal adjustment ability for target state tracking with optimal control action behavior.It can also overcome the steady-state error caused by model parameter mismatch and external disturbance,and achieves optimal control of speed and current.