Based On The Extended Kalman Filter AC Permanent Magnet Synchronous Motor Parameter Identification

Sensorless control technique based on state estimation and parameter identification has become the new research focus and development trend of permanent magnet synchronous motor (PMSM) control. The influence of measurement noise on motor dynamics can be effectively suppressed or reduced by replacing the position sensor with certain control algorithms. Therefore, the cost performance of control system is improved. Currently, the approaches of sensorless control are mainly classified as back electromotive force, model reference adaptive system, extended Kalman filter (EKF), etc.As an optimal recursive algorithm for nonlinear system, EKF is able to process the data with model error and measurement error, which means the filter has excellent ability of adaption and robustness. PMSM control system based on EKF can track the variation of motor parameters in real-time and the estimation results can be treated as on-line adjustment reference of controller parameters. Consequently, fast dynamic performance and good dynamic quality of the control system are derived. In addition, using EKF algorithm to estimate rotor position of PMSM takes the advantage of initial position irrelevance, which simplifies start-up process of motor and makes EKF a prospective alternative of sensorless control schemes.All the research work are focused on surface-mounted AC PMSM in this thesis. At the beginning, control technique of PMSM and parameter identification approaches for motors are briefly introduced. Subsequently, mathematical models of PMSM in different reference frames are given. Meanwhile, vector control technique and hysteresis current control of voltage source PWM inverter are also discussed. Then, the theory of Kalman filter is studied and analyzed in detail, including linear Kalman filter algorithm and EKF algorithm. Based on things mentioned above, the identification models of stator resistance, rotor flux-linkage, rotor position and speed, load torque as well as temperature of stator and rotor are investigated, while the concentration is the estimation performance of reduced-order models. In the end, the vector control system of PMSM and parameter identification models are established in MATLAB/SIMULINK, and explicit simulation analysis of performance of different models are carried out. Simulation results demonstrate that EKF algorithm can track the variation of motor parameters precisely and fast, while the EKF observer involved vector control system of AC PMSM features good performance.