Calibration of Torque Error of Permanent Magnet Synchronous Motor Base on Polynomial Linear Regression Model

Abstract

Permanent magnet synchronous motor is widely used in various fields. However, there is a nonlinear torque error between measured torque and calculated torque because of the nonlinear parameters variation of the motor. On one hand, in order to reduce the nonlinear torque error, a general calibration method of torque error based on polynomial linear regression model and a torque closed-loop control strategy based on the above calibration method are implemented. On the other hand, in order to find an optimal regression model, both the univariate regression model on different orders and the bivariate regression model on different orders are presented and analyzed through comparison criteria. The comparison results show that the torque error of univariate regression model is larger than that of bivariate regression model and bivariate second-order polynomial linear regression model is the optimal one. Furthermore, according to the comparison results, the calibration method based on bivariate second-order polynomial linear regression model is obtained and then applied in torque closed-loop control strategy to achieve accurate motor control. Experimental results of the motor control strategy in this paper show that the control performance is with a good torque accuracy and with a stability in the whole ranges of speed and torque.