Parameter estimation for VSI-fed PMSM based on a dynamic PSO with learning strategies

A dynamic particle swarm optimization with learning strategy (DPSO-LS) is proposed for key parameter estimation for permanent magnet synchronous machines (PMSMs), where the voltage-source-inverter (VSI) nonlinearities are taken into account in the parameter estimation model and can be estimated simultaneously with other machine parameters. In the DPSO-LS algorithm, a novel movement modification equation with variable exploration vector is designed to effectively update particles, enabling swarms to cover large areas of search space with large probability and thus the global search ability is enhanced. Moreover, a Gaussian-distribution based dynamic opposition-based learning (OBL) strategy is developed to help the pBest jump out local optima. The proposed DPSO-LS can significantly enhance the estimator model accuracy and dynamic performance. Finally, the proposed algorithm is applied to multiple parameter estimation including the VSI nonlinearities of a PMSM. The performance of DPSO-LS is compared with several existing PSO algorithms, and the comparison results show that the proposed parameters estimation method has better performance in tracking the variation of machine parameters effectively and estimating the VSI nonlinearities under different operation conditions.