抄録
Product configuration optimization at early design stages requires the simultaneous evaluation of numerous design aspects. However, the existence of complex relationships among design alternatives, such as subordinate and constraint relationships, increases the difficulty of formulating the problem model and conducting straightforward searches for optimal solutions. This study presents a method for representing design alternatives having such relationships that uses hierarchical genotypes to enhance the design optimization efficiency of genetic algorithms. Furthermore, the efficiency of different genotype structures is compared using numerical experiments. Based on the experimental results, a procedure for constructing optimal hierarchical genotypes is proposed. The proposed methods are then applied to a robot arm design optimization problem to illustrate their effectiveness.
