Mechanical Parameters Identification And Disturbance Rejection Control For Permanent Magnet Synchronous Motor Drive Systems

The disturbance caused by the variations of load torque and inertia will degrade dynamic performance of permanent magnet synchronous motor drive system. In order to improve the ability of disturbance rejection for PMSM drive system, the value of system disturbance should be identified accurately and expeditiously, which will be then compensated in controllers by some agreed anti-disturbance strategy. The identification algorithms for inertia, load torque and system anti-disturbance strategy are researched in this paper. The parameters coupling and the identification fluctuation problems in existed methods are analyzed. On this basis, a mechanical parameters identification algorithm based on DOB is proposed.By choosing speed and current closed loop vector control system as object, the recursive mechanism of traditional mechanical parameters identification is firstly established in this paper, and then the existence of coupling problem is analyzed. To solve the coupling problem, a novel mechanical parameters identification algorithm based on DOB is designed by using Landau adaptive theory. The algorithm takes mechanical system equation as the reference model, reduces the influence due to the coupling problem by choosing high independent parameters to be identified which include the information of inertia and load torque. At the same time, the identification fluctuation when parameters change is analyzed and depressed by choosing reasonable adaptive gains. At last, an anti-disturbance control strategy with PI tuning and load feed forward compensation is proposed so as to improve the ability of disturbance rejection for PMSM drive system.The Landau adaptive identification strategy is compared to traditional strategy by the experiment on PMSM experimental platform. The experimental results show that the proposed algorithm can realize the mechanical parameter identification quickly and accurately, and the identification result fluctuation when parameter changes is depressed effectively.