Model Reference Adaptive Observer for Permanent Magnet Synchronous Motors Based on Improved Linear Dead-Time Compensation

Abstract

Aiming at the problem that the model reference adaptive observer (MRAS) is sensitive to the parameters of the motor model, this paper designs a model reference adaptive observer with resistance adaption and considers the influence of the inverter dead zone. The method introduces the online identification of resistance on the traditional reference adaptive observer model, corrects the resistance parameters of the model’s reference adaptive observer in real time, selects PI as the adaptive law, and proves the stability of the adaptive law using Popov’s stability theorem. To study the influence of the inverter dead zone on the voltage parameters in the estimation model at low speeds, an improved linear dead zone compensation method is proposed to improve the model reference adaptive observer, which eliminates the voltage error between the estimated motor model and the actual motor model. For a given rotational speed of 300 rpm, the observed errors of sliding mode observer (SMO) and traditional MRAS in a steady state were 8.3% and 6%, respectively, and the rotational speed error was controlled to 1.6% using the compensation scheme proposed in this paper. After compensation, the resistance identification error was reduced from 17% to 1.6%. The simulation and experimental results showed that the accuracy of position estimation can be significantly improved using dead zone compensation under low-speed conditions, and the dead zone compensation can improve the accuracy of the online recognition of the resistance.