Research On The High-performance Sensorless Control Of PMSM

Because of several advantages, such as compactness, high power density, high efficiencyand good operation performance, permanent magnet synchronous motors (PMSMs) areattracting extensive attention in theory research and practical application. Now, PMSMs,which are used as a core implementation component, have been widely used in CNC, robot,aerospace, marine and other fields. In this paper, according to the requirements of the electricvehicle control system, this dissertation made the studies on sensorless control of PMSM.Taking the sensorless PMSM control system as the research target, this dissertation conductedsome research in-depth on the following issues, including on-line estimation of the rotorposition and speed, control method of the motor, dead-time compensation and so on. It aimsto reduce the cost and complexity of drive system, and to further improve the reliability andcontrol performance of motor control system.Based on coordinate transformation, PMSM model in the two-phase frame system wasderived firstly, and then an Extended Kalman Filter (EKF) for closed-loop rotor speed andposition estimation of PMSM was designed based on the above model. Although the initialrotor position is unknown, the sensorless control stategy has key ability of self-startup.Whereas, during startup transient, convergence error and stall issues may happen. In order tosolve these problems, detailed explanations for these issues are given and relevant solutionsare discussed.Considering that the noise covariance matrix has a great impact on estimationperformance, the influence on estimates by noise covariance matrix is analyzed by simulationresults, and the trial-and-error method for setting noise covariance matrix is summarized.This extended Kalman filtering technique requires complete specifications of dynamicalmodel parameters to guarantee estimation accuracy, and impact on estimation accuracy byparameters variation is summarized by simulation results. To solve this problem, an in-depthstudy on adaptive fading extended Kalman filter (AFEKF) is made. A fading factor, whichenhances the influence of innovation information, may be incorporated as a multiplier forimproving the tracking capability and estimation accuracy in high dynamic control of PMSM.To reduce computational complexity, a nonlinear two-stage extended Kalman filter(NTSEKF), which employs the two-stage structure, is proposed by decoupling the EKF intotwo parallel reduced-order filters. By using the number of arithmetic operations(multiplications and additions) as the measure of computational complexity, thecomputational advantage of the two-stage Kalman filter over the conventional Kalman Filter has been demonstrated. Because EKF is uniformly asymptotically stable, the stability ofNTSEKF is verified by showing that NTSEKF is equivalent to EKF. Combining advantagesof NTSEKF and AFEKF, adaptive two-stage extended Kalman filter (ATEKF) is developedby decoupling the AFEKF into two parallel reduced-order filters. ATEKF has both strongrobustness against model-plant parameter mismatches and good real-time state tracking ability.The stability of ATEKF is verified by the same method as in NTSEKF.Currently, linear PI controllers were mostly employed for the control of the current loop.PI controller gains excellent steady-state performance, but the dynamic response is slow andthe control effect depends greatly on the PI parameters. In order to overcome the drawbacksof the PI controller, this paper presented a novel control scheme for current loop by replacingthe PI controllers with predictive current control (PCC) controller. Inaccuracy in systemmodels may seriously degrade the performance of the PCC controller. Treating inaccuraciesas disturbances, the disturbances caused by the parameter variations and dead time arecompensated by the two online observers, respectively. Aiming at the saturation nonlinearityof the speed control loop PI controller, a variable structure anti-windup PI controller isproposed to improve the speed control performance.Based on Expert3system, a full-digital sensorless PMSM control system is designed torealize the ATEKF and PCC controller. Based on these algorithms and hardware, in-depthsimulation research and experiment validation are made.