A Method for Shape and Topology Optimization of Mechanical Structures Using Genetic Algorithm. 2nd Report. On Convergence of Solutions of Our Method by Adoption of Removal and Addition Parameters of Elements as Chromosomes.

Abstract

In the previous paper we proposed a method for shape and topology optimization (i. e., layout optimization) of mechanical structures using a genetic algorithm. The method is contrived in such a way that the behaviors of layouts are investigated to obtain the optimum layouts by selecting the layouts of mechanical structures as individuals, and by adopting removal and addition parameters of elements as chromosomes. It was demonstrated by various experiments that the method is efficient and stable in convergence of the solutions. In this paper good convergence of the method is confirmed by schema theorem and experiments. Several strategies are also proposed, such as scaling, elite strategy, and our method for limiting the search domain. For the purpose of improving the convergence. the effects of these strategies on efficiency and stability of convergence are discussed together with the results of experiments.