Multi-level Thrust Ripples Minimization of Linear Flux Switching Motors With Segmented Secondary by Combined Genetic Algorithm and Response Surface Methodology

• Published in: Iranian journal of electrical & electronic engineering Vol. 16; no. 1; pp. 96 - 106
• Main Authors: M. S. Hosseini, H. Javadi, S. Vaez-Zadeh
• Format: Journal Article
• Language: English
• Published: Iran University of Science and Technology 01-03-2020
• Subjects: linear flux switching motors; multi-level optimization; response surface methodology; sensitivity analysis method; thrust ripples minimization
• ISSN: 1735-2827; 2383-3890
• DOI: 10.22068/IJEEE.16.1.96

Linear flux switching motors with simple passive segmented secondary, referred as Segmented Secondary Linear Flux Switching Motors (SSLFSMs), have low cost secondary and therefore are applicable to transportation systems like Maglev. However, it is shown that the SSLFSMs suffer from high thrust ripples. In this paper, minimizing SSLFSM thrust ripples besides maximizing its developed thrust are performed by considering the motor dimensions as design variables. Since the optimization of the motor is a high dimensional problem, a multi-level optimization method is employed to improve the machine performances and efficiency. According to the effects of the design variables on the optimization objectives, a sensitivity analysis is carried out to divide the design variables into two levels: mild-sensitive level and strong-sensitive level. Then, the two levels of design variables are optimized based on a mathematical model. Two different optimization methods as the Design of Experiment (DOE) and the Response Surface Method (RSM) are used in mild-sensitive level and the Genetic Algorithm (GA) is also used in strong-sensitive level. Based on FEM analysis, electromagnetic performance of the original motor and the optimal one are compared and the validity of the proposed optimization method is verified. Also, the effectiveness of the mathematical model used in thrust and thrust ripples calculations is evaluated and verified.