遺伝的アルゴリズムをベースとした繊維配向角の最適化手法の提案

抄録

This study proposes a genetic algorithm (GA)-based optimization method for fiber orientation angle and examines its effectivity. Determining the optimal fiber orientation angle involves searching for a solution from multitudinous combinations, which makes it difficult to attain the optimal solution by simple GA. In addition, the generally used principal stress design cannot give an optimal solution when combined loads are applied. Therefore, an optimization method is developed that does not directly treat the fiber orientation angle as a design variable and attains the optimal solution with lesser design candidates than simple GA. Herein, stiffness maximization is formulated as a strain energy minimization problem, and optimization is performed using the proposed method, wherein the orientation angle of each element of the target FEM model is treated as a design variable. The resultant design candidate (with the flat plate and B-pillar models under combined loads) optimized by the proposed method yields a higher stiffness than that optimized by its conventional counterparts. Thus, it is shown that the proposed method obtains a better design candidate than that obtained by simple GA and principal stress design methods.