Dynamic Multilevel Optimization of Machine Design and Control Parameters Based on Correlation Analysis

In this paper, a multilevel optimization method is proposed for a motor drive system including a surface-mounted permanent magnet synchronous machine (SPMSM), the converter/inverter, and the control schemes. First, the multilevel optimization is described by using the problem matrix which may be used to allocate the design variables on different levels. The parameters in the problem matrix are deduced by using correlation analysis. Second, the architecture and implementation of multilevel genetic algorithm (MLGA) are carried out. As one of the advantages of MLGA, the dynamic adjustment strategy of GA operators is utilized to improve the optimal performance. The algorithm is then applied to a three-level optimization problem in which the optimization of SPMSM design and the control parameters of drive are considered in different levels. Finally, some results and discussions about the application of the proposed algorithm are presented.