論文ID: 25-00086
Finite elements analysis (FEA) has been widely used for evaluating the performance of products in practical engineering design processes. In the FEA of industrial equipment, such as large-scale motors and transformers, the accurate identification of material constants for iron core components is essential. This identification is formulated as an inverse problem based on mechanical properties. Surrogate multi-objective optimization has been widely employed to solve this problem, demonstrating high accuracy. However, the identified material constants are not always uniquely determined and may result in physically unrealistic values. From a practical perspective, it is crucial to obtain physically realistic material constants. To address this issue, this paper presents two key ideas: (1) the introduction of an in-plane isotropy constraint that reflects the structural characteristics of the iron core, and (2) the consideration of the sensitivity of objective functions with respect to material constants. The former effectively confines the search space to a physically plausible region, thereby enhancing both the physical realism of the identified constants and the accuracy of the FEA results. The latter contributes to improving the uniqueness of the identified values. These ideas are validated through a case study of an iron core component. The results of this study show the effectiveness in improving identification accuracy and ensuring the physical plausibility of the estimated material constants.