Global sizing optimisation using dual‐level response surface method based on mixed‐resolution central composite design for permanent magnet synchronous generators

Abstract

This study presents a global sizing design optimisation of a permanent magnet synchronous generator (PMSG) using the three-dimensional finite-element analysis (3D FEA). To build an optimal parametric model structure, the efficiency improvement of the PMSG is taken as the main objective, where iron and copper losses were minimised. A dual-level response surface methodology (D-RSM) with a window-zoom-in approach for a variable-speed-range analysis as a global optimisation technique is employed to find out the optimal design variables of the objective function. The D-RSM using mixed-resolution central composite design (MR-CCD), full factorial design, central composite design (CCD), and box-Behnken design are applied to optimise the geometry with very small error. Analysis of variance and multi-level RSM plots are used to check the adequacy of fit. However, the MR-CCD exceeds the range of the boundary in the design region. Hence, a modified MR-CCD is used that improves the efficiency and proposes the parameter settings to manufacture the high-class quality wind generator. The validation of the analytical and numerical fashions is successfully achieved through rigorous FEA, and the experimental verifications perfectly marked the theoretical and significance optimisation design.