Particle Swarm OptimizationとSLPのハイブリッド化による機械システムの多目的最適化(機械要素,潤滑,工作,生産管理など)

概要

論文の詳細を見る
Multiobjective optimization techniques such as Evolutionary Algorithms, which can obtain Pareto optimal solution set in a single calculation, have attracted attentions to conduct trade-off analysis in mechanical design process. Multiobjective Particle Swarm Optimization (MOPSO) has an advantage in obtaining Pareto optimal solution set in structural optimization problem since it can easily handle continuous design variables. However, this algorithm has not been applied to complicated structural and mechanical optimization problems due to its difficulty of constraint handling in its local search process while providing high-performance in global search. This paper proposes an approach to obtain Pareto optimal solutions using MOPSO combined with the Sequential Linear Programming (SLP) to enhance local search in multiobjective optimization. The MOPSO update scheme is applied to dominated solutions and the SLP is applied to non-dominated solutions. The application of the SLP along with optimal weighting coefficients results in a stable and fast in multiobjective optimization. Finally, truss structure design optimization problems are solved to verify efficiency of the proposed algorithm.
社団法人日本機械学会の論文
2009-04-25