An Efficient Hybrid Algorithm for Optimization of Discrete Structures

Abstract

Presented in this paper is a hybrid algorithm for the design of discrete structures like trusses. The proposed algorithm called Discrete Structures Optimization (DSO) is based on the Evolutionary Structural Optimization (ESO) [1,2]. In DSO, material is removed from the structural elements based on the strain energy. DSO is a two stage process. First stage is the topology optimization where the elements of the structure with the least amount of strain energy are identified and eliminated. The second stage is the sizing optimization of the structure with optimum topology identified in first stage. For the continuous design variables a gradient based method is used and for the discrete design variables a genetic algorithm is used. The algorithm is tested on 2-D and 3-D discrete structures. DSO results show significant reduction in the number of finite element analysis (FEA) evaluations as compared to genetic algorithms using simultaneous topology and sizing optimization.KeywordsGenetic AlgorithmTopology OptimizationHybrid AlgorithmTruss StructureTopology Optimization ProblemThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.