Comparison of Online Inertia Identification Methods for Permanent Magnet Synchronous Motor

Abstract

For achieving high performance of servo drives, the inertia has to be estimated with an online identification algorithm, especially if the inertia is likely to change due to the drive application, such as robots. Suitable identification method can be chosen more proficiently, if their benefits and shortcomings are described clearly. In this paper, three common inertia identification methods for PMSM are compared with regard to their identification accuracy, convergence and computation time in a MCU: the recursive least squares (RLS), the model reference adaptive control (MRAC) and the extended Kalman filter (EKF). As a result, the RLS was selected and implemented for an existing servo drive system.