抄録
The advantage of the carbon fiber reinforced plastics (CFRP) is tailoring capability for their mechanical properties by changing the lay-up designs. Recently, meta-heuristic methods (e.g., a genetic algorithm (GA) and a particle swarm optimization (PSO)) have been often employed in such lay-up design problems. In addition to the proper choice among meta-heuristic methods, there are some parameters to be decided carefully to achieve good performance in optimization. In practical problems involving heavy computation costs, however, it is difficult to judge which method is appropriate to a given specific problem and what values are reasonable for the accompanying parameters. In this study, a new benchmark function is proposed to simulate global behaviors of the lay-up design problem and to give criterion for proper choice of the parameters in GA, PSO and the layerwise optimization (LO) methods. In numerical results, performances of the three methods are compared and the proposed benchmark function is shown to be quite effective in simulating the lay-up optimization problem.
