412 モードトリミング法における局所的変数変換によるカオス挙動と大域的探索能力(GS-12 設計工学,研究発表講演)

概要

論文の詳細を見る
A gradient-based method for global optimization of nonlinear programming problems, named "Modal Trimming Method," has been proposed. It has turned out that a hybrid method which combines an extended Newton-Raphson method with an extended secant method promotes a chaotic behavior and avoids periodically-vibrating behaviors in renewing the values of variables to search feasible solutions, which can enhance the global searching capability and computational efficiency. In this paper, a local variable transformation is proposed to attain further enhancement of the computational efficiency. It is shown that the possibility of staying in the regions of local optimal solutions found previously is lowered, and that of finding better local optimal solutions with fewer iterations for the renewal of the values ofvariables is heightened.
一般社団法人日本機械学会の論文
2005-03-18