Research On Position Sensorless PMSM System Based On Extended Kalman Filter

Because the incomparable advantages, such as high power density, high efficiency, high performance and so on, permanent magnetic synchronous motor (PMSM) has been theoretically studied deeply and used widely in industrial application. PMSM and its control system, has also been used in CNC, robot, aerospace, marine and other fields. In some application fields, the reliability of the control system is the most important. According to the requirements of the submarine control system, this dissertation made the studies on sensorless control of PMSM. It aims to improve the reliability of motor control system, by solving the seal problems of traditional transducers, and with the hope to gain some theoretical breakthrough in the research area of the PMSM control system.Taking the sensorless PMSM control system as the research target, this dissertation conducted some research in-depth on the following issues, including the detection of the rotor position and speed, on-line estimation of motor parameters, control method of the motor and so on.In this paper, PMSM model at the two-phase stationary frame system was derived firstly, and then an Extended Kalman Filter (EKF) was designed based on the models. In the modeling of PMSM at the two-phase stationary frame system, the influence of inertia on the filter tracking performance was considered. It is difficult to select the EKF matrix parameters which are different in different motor systems, which prevents EKF from being applied in the position sensorless PMSM system. To solve this problem, the normalization parameter is used in this dissertation to make the matrix parameters uniformity; which brought the benefits including the reduction of the system developing time and great perspective in the practical application.When using EKF to estimate rotor position and speed, random parameters would affect estimated results. This paper conducted researches on the influence of model parameters and system noise on the filter estimation and validated the results by making simulations and experiments. Different filter parameters appeared in dynamic or static process would affect estimated results, so variable parameter estimation was presented, in order to guarantee the filter performance, and meet the dynamic and static performance demand of the system.The variation of Motor parameters has great influence on EKF performance. So to study the on-line identification of parameters in depth is necessary. In this dissertation, model referenced adaptive identification is adopted. Based on Popov super-stabilization theory together with motor models, a double-structure identification model is designed, to identify motor resistance, inductance and rotor flux on-line, in order to compensate EKF estimation precision effectively.In order to improve the system performance of dynamic response and disturbance rejection, an integral sliding mode variable structure control method is used to control the speed of PMSM, which is robust enough under external disturbances and the variation of system parameters. In the actual applications, sliding mode variable structure control system would appear chattering due to sampling time and other factors. To reduce chattering brought by variable structure control, a combined control method is provided, based on fuzzy slidling mode variable structure control.A double-DSP full-digital sensorless PMSM control system is designed to realize the above algorithms. SPI interface is used in the system as the communication standard between two DSPs, to communicate data in real-time. Based on these algorithms and hardware, in-depth simulation research and experiment validation are made.