This article is dedicated to achieving the best thrust performance of high thrust and low thrust ripple of permanent magnet synchronous linear motors (PMSLMs), which are used in precision positioning system. To achieve this, a novel PMSLM with double-layer reverse skewed coil (DRSC) is designed and optimized. First, the topological structure of the DRSC-PMSLM is introduced, and the effectiveness of the DRSC is analyzed qualitatively according to the analytic model. Certain hypotheses that are more suited to qualitative analysis can lead to the low accuracy of the analytic model. Therefore, for the subsequent optimization, the weighted random forest (WRF), an enhanced random forest (RF) algorithm by assigning different weights to different regression trees, is proposed to fit the sample data generated by the finite element method to establish a high-precision proxy model. Accuracy test proves that the WRF model has higher accuracy than the standard RF model and the analytical model. Subsequently, a new global optimization algorithm, call the modified Krill Herd (MKH) algorithm is proposed to iteratively optimize the WRF model to obtain the optimal structure parameters. The MKH algorithm is a modified KH algorithm that two strategies are used to improve the convergence rate and avoid premature convergence. Comparative experiments are performed by using genetic algorithm, particle swarm optimization, and standard Krill herd, which prove that the MKH has faster convergence speed and stronger global search capability. The thrust performance of the optimized motor is improved considerably compared with that of the initial motor, which further proves the effectiveness of the MKH used in this article. Finally, the prototype experiment proves that the PMSLM with DRSC has the best thrust performance.
Read full abstract