Simultaneous size and shape structural optimization using enhanced comprehensive learning particle swarm optimization

Abstract

This study proposes a simultaneous size and shape structural optimization on truss structures using a metaheuristic algorithm, namely Enhanced Comprehensive Learning Particle Swarm Optimization (ECLPSO) which is based on the CLPSO algorithm. CLPSO itself was originated from a famous Particle Swarm Optimization (PSO) that was invented based on behaviour or movement of a bird flock. CLPSO is an improved PSO algorithm which has good exploration ability but is poor in exploitation. Thus, ECLPSO introduces of two enhancements, including perturbation-based exploitation and adaptive learning probabilities, to improve the exploitation ability and to adjust the learning probabilities of each particle, respectively. The objective of the study is to obtain the minimum weight of the benchmark truss structures under the design constraints required specifically for each problem. By combining size variables and shape variables of the problems, the optimizations shall produce better results than those of size optimizations alone. In this research, the optimizations are implemented in Python code and performed several times individually to conclude the robustness and accuracy of the algorithm. The results including the minimum weight and standard deviation of the solutions shall be compared and discussed with those of some other metaheuristic algorithms in the literatures.