Inverse Shape Optimization Using Dynamically Adjustable Genetic Algorithms.pdfVIP

Inverse Shap e Optimization Using Dynamically Adjustable Genetic Algorithms  Vlatko Cingoski, Memb er, IEEE Norio Kowata, Non Memb er Kazufumi Kaneda, Non Memb er Hideo Yamashita, Memb er, IEEE Hiroshima University Toa Medical Electronics Co., LTD, Higashihiroshima, 739, JAPAN Kob e, JAPAN A bstract| In this pap er, a new dynamically adjustable ge- Genetic algorithms b elong to the group of probabilis- netic algorithm for inverse shape optimization of electrical tic searching metho ds and they have a high probability devices is proposed. The algorithm starts with initial p op- ulation which is not entirely randomly dened and dynam- of locating the global optimum in the multidimensional ically changes the p osition and the width of the searching searching space discarding all existing local optima. The space as the searching pro cedure evolves with time and the implementation of GAs for inverse optimization of electro- ob jective function approaches its optimum. The prop osed magnetic devices has b een recen tly prop osed [2], [3]. The algorithm is successfully applied for inverse shape optimiza- tion of a die mold press machine and for pole shap e opti- robustness of the metho d, its ability to deal directly with mization of a rotating machine. To achieve smo oth pole face optimization variables without any derivatives and usage the optimized shape is dened using several control p oints of an enco ded binary represen tation of the optimization and ordinary spline functions.