Online Identification of PMSM Parameters: Parameter Identifiability and Estimator Comparative Study

In this paper, a model-reference-based online identification method is proposed to estimate permanent-magnet synchronous machine (PMSM) parameters during transients and in steady state. It is shown that all parameters are not identifiable in steady state and a selection has to be made according to the user's objectives. Then, large signal convergence of the estimated parameters is analyzed using the second method of Lyapunov and the singular perturbations theory. It is illustrated that this method may be applied with a decoupling control technique that improves convergence dynamics and overall system stability. This method is compared with an extended Kalman filter (EKF)-based online identification approach, and it is shown that, in spite of its implementation complexity with respect to the proposed method, EKF does not give better results than the proposed method. It is also shown that the use of a simple PMSM model makes estimated parameters sensitive to those supposed to be known whatever the estimator is (both the proposed method and EKF). The simulation results as well as the experimental ones, implemented on a nonsalient pole PMSM, illustrate the validity of the analytic approach and confirm the same conclusions.