領域遺伝型遺伝的アルゴリズムの開発 : 領域の設定方法(機械力学,計測,自動制御)

概要

論文の詳細を見る
As the most of the problems is engineering design are the mixed variables with discrete and continuous variables, Gas is thought as a suitable tool for solving these difficulties. However, Gas has a problem in expressing continuous variables ; therefore it is not as effective as they are thought. There are a number of studies in Gas that treat expression of continuous variables. Most of them did not use G-type expression to make sure accuracy of variables. Indeed, it seemed difficult to make sure accuracy of variables by using binary expression of G-type expression, we have developed adaptive range genetic algorithms, which change searching range for each generation by using information of convergence and solve these problems. However, searching range changes each generation, so that it does not have stability in searching. Therefore we needed to introduce some system parameters and rules to keep it stable. In the previous study, we introduce a new concept of inheritance of searching range and get rid of the stability of searching range problem. In this study, we propose a new method of setting searching range so that it has both minute search to raise accuracy and global search for global optimization.
社団法人日本機械学会の論文
2003-10-25