Parameter-tuning-free two-step identification of mechanical parameters for PMSM drives

Abstract

Accurate identification of mechanical parameters holds great significance for optimizing control performance and monitoring working status in permanent magnet synchronous motor (PMSM) drives. This work proposes a parameter-tuning-free two-step mechanical parameter identification method using signal injection and algebraic operations. It is devised based on a nonlumped mechanical motion model that incorporates nonlinear and asymmetric friction modeling, which can accurately characterize real-world dynamics. More importantly, the proposed method circumvents the need for parameter adjustments, such as observer gain design and pole placement. With a two-step collaborative mechanism, the developed method is successfully implemented. In the first step, the open-loop speed response triggered by signal injection is exploited to accurately identify nonlinear friction along with the initial inertia value. With aiding from the results of the first step, the second step develops an online algebraic estimator to track the potentially time-varying load torque and inertia. Real-time experiments are conducted on a 1.2-kW practical PMSM drive system to validate the feasibility of the presented method.