2307 ハイブリッド型多目的PSOを利用した信頼性に基づく多目的最適設計(OS6-2 設計と最適化II-多目的最適化-,OS6 設計と最適化)

概要

論文の詳細を見る
This study proposes a reliability-based multiobjective optimization (RBMO) algorithm incorporated with hybrid multiobjective particle swarm optimization (MOPSO) handles the constraint satisfaction technique based on gradient information of constraint conditions and the single loop reliability-based optimization techniques, single-loop-single-vector method (SLSV). The authors have been proposed a hybrid MOPSO for handling the constrained multiobjective optimization. In the algorithm, the violated particle will move to feasible region using gradients of the constraint conditions. In this study, the hybrid MOPSO is extended to the reliability-based multiobjective optimization that the reliability is usually evaluated by using the gradient information of the limit state function such as the first order reliability method (FORM). However, the FORM-based optimization approach was formulated as double-loop iteration method. Therefore, SLSV method formulated as a single-loop method by using a proper approximation of the reliability constraint is adopted for computational efficiency. Through numerical examples, the effectiveness of the proposed method is demonstrated.
一般社団法人日本機械学会の論文
2011-10-21