PMSM Design optimization concerning sensorless performance and torque ripple

Abstract

This paper proposes a multi objective design optimization approach concerning sensorless performance and torque ripple for Permanent Magnet Synchronous Machine. Sensitivity analysis of design parameters has been accomplished for critical parameters selection and metamodels are generated based on critical variables. Based on these metamodels, the optimization process aims to propose a machine geometry to improve sensorless performance without compromising torque ripple and average torque. The metamodel assisted optimization process has been proved to reduce computational cost with a good accuracy. The selection of the optimal solution was achieved based on multi-objective optimization process using NSGA-II. The optimal solution is validated by finite element analysis and has proved an extending of 100% of the feasibility area of sensorless control with only an increase of 3% of torque ripple.