Position sensorless control of PMSM under low-frequency ratio

Abstract

In conventional approaches to sensorless PMSM control, Luenberger observer designs are commonly developed and evaluated within the continuous-time domain, with subsequent discretization typically implemented through the first-order forward Euler approximation method. However, when the ratio of the switching frequency to the motor operating frequency is relatively low, traditional observer design methods face issues where current estimation errors fail to converge rapidly or even become unstable. To address the inaccuracies of the Euler discretization method, this paper first establishes an accurate discrete-time mathematical model for single-sampling and single-update PMSM drive systems. Secondly, to simplify parameter calculations and address the computational complexity arising from the asymmetry of the inductance matrix in the mathematical model of salient-pole motors, an approximate symmetry approach is proposed. Based on an improved mathematical model, a full-order back-EMF observer is established to achieve sensorless control of PMSMs. Finally, to mitigate the impact of current mean sampling errors on the position observer, a compensation measure is proposed. Simulation experiments conducted in MATLAB/Simulink validate the proposed methods, and the results verify that the approach performs as expected.