Online Multiparameter Identification Method for Sensorless Control of SPMSM

The performance of the model-based sensorless technique depends on the accuracy of model parameters. Parameter errors between the model and the machine may result in the estimation error of the rotor position, and then degrade sensorless control performance. This article proposes an online multiparameter identification method, which is able to estimate the resistance error, inductance error, permanent magnet flux linkage, and the equivalent amplitude of the voltage distortion caused by inverter nonlinearity. Data required by the proposed identification method are obtained by injecting current pulses into the estimated d-axis. Estimating machine parameters without the help of position sensors requires solving a multivariate nonlinear regression problem, which is different from parameter identification when position sensors provide a rotor position. This article analyzes the rank-deficient problem of the proposed parameter identification method and introduces conditions that the currents in the estimated dq-axis need to meet when the rotor speed is constant. The effectiveness of the proposed parameter identification method is verified by simulation and experiment.