Educated Reducing the effect of GA parameters on optimization of Topology and cross section for truss structures using Multi-Search Method

Abstract

Although GA optimization is among optimization methods, the method may not be successfully employed in all cases due to slow down in process and some other unknown parameters including the number of generations, cross over ratio, mutation ratio, and the selection process which cause local optimized points. In this paper a new approach is proposed to perform GA for optimization of cross section and topology of trusses that reduces such problems. A complete system with different sub sections, called island, is used to search in the design space. In each island, different operators and parameters are used separately. After some generations, depending on the migration ratio, the best chromosomes from each island alter the chromosomes with lower fitness in other island. Based on the proposed method, GA is continued until the global optimum with the least dependence on the GA parameters is achieved. The results were evaluated with some standard examples.

Keywords

• costing
• genetic algorithm
• truss structures
• topology
• multipurpose search method