Evolutionary algorithm using surrogate assisted model for simultaneous design optimization benchmark problem of multiple car structures

This paper proposes a surrogate-assisted evolutionary algorithm for solving optimization problems with high calculation cost for constraint determination. The proposed method consists of CMOEA/D that extends the ability of MOEA/D to deal with constrained optimization problems and a surrogate evaluation model constructed by a machine learning, extreme learning machine (ELM). To investigate the effectiveness of the proposed method, we conduct an experiment on simultaneous design optimization benchmark problem of multiple car structures developed by Mazda Motor Corporation et al.. The experimental result revealed that the proposed method can obtain optimal solutions faster than CMOEA/D without a surrogate model.