差分進化に基づくセットベース設計手法とマウント設計への応用

抄録

In the vehicle development, it is important to solve a feasible design set where all multidisciplinary constraints are satisfied. By solving feasible design set instead of a single optimal design, critical rework due to uncertainties of design conditions can be avoidable. In this paper, we propose a novel set-based method based on Differential Evolution Based Adaptive Sampling (DEBAS) which can solve feasible design set of an optimization problem with high-dimensional design variables. In the proposed adaptive sampling strategy, the mutation scheme in differential evolution (DE) is introduced. For objective function used in the DEBAS, Kreisselmeier-Steinhauser function is introduced to evaluate dominant violation value among constraint functions. Besides, the objective function has a term which can generate samples throughout feasible region. To show the effectiveness of the proposed method numerical examples including practical design problem are demonstrated.