An Improved Extended State Observer Based on Linear-Nonlinear Switching Strategy for PMSM Sensorless Control

The Observer based sensorless control method is very popular and practical in the PMSM sensorless control field. The angle and speed estimation scheme of a typical sensorless control method includes two major part: the back-EMF observer and the angle-speed estimator. In this paper, the traditional observer based sensorless control method is introduced and analyzed. Then, an extended state observer (ESO) based back-EMF estimation solution is proposed, which reconstructs the position information by formulating the back-EMF information as states variable and estimating them online with the observer. Considering that the efficiency of suppressing estimation error is high dependent on the ESO's feedback mechanism, a linear-nonlinear switching strategy is proposed and analyzed to improve existing feedback function. The proposed method not only combines the linear ESO's good tolerance to the disturbance and the nonlinear ESO's high estimation accuracy, but also reduces the nonlinear ESO's sensitivity on parameters and makes it easier to use. As for angle and speed estimation, the traditional PLL method brings ripple and phase delay to the estimated speed. In this paper, an ESO based angle and speed observer is incorporated into the PLL to improve the speed estimation accuracy. Simulation result proves the advantage of the proposed method over the traditional observer-based sensorless control method.