This article presents a multiobjective robust optimization design method for accelerating the transition from design to application of low-volume manufacturing and customized machines. This study primarily considers the impact of manufacturing tolerances on a novel hybrid-excited generator with stator slot permanent magnets. Initially, optimization objectives are proposed based on the application and structural characteristics of the machine. Then, sequential sampling is introduced to increase modeling samples gradually, synergizing with the local outlier factor (LOF) to establish an enhanced radial basis function neural network (RBFNN) surrogate model. Furthermore, three optimization methods are comprehensively compared: deterministic optimization algorithm, robust experimental design, and robust optimization algorithm. Monte Carlo simulation (MCS) is used to evaluate the robustness of the results obtained by the three methods. The comparative analysis reveals that the robust optimization algorithm produces a more concentrated performance distribution and exhibits smaller result fluctuations in the context of manufacturing tolerances. Finally, the effectiveness of the proposed robust method is verified through finite-element model analysis, and a prototype is manufactured and tested to validate the simulation results.
No takes yet. Share an insight, caveat, or question.
Zhao et al. (2024) studied this question.