Design procedure for laminated composites based on genetic algorithm with function approximation
Zako Masaru; Takano Naoki; Takeda Norio
Abstract:In this paper, a design procedure for laminated composite structures based on GA (Genetic Algorithms) with function approximations is proposed. Design variables considered to enhance the rigidity of the laminated composite structure are the number of laminate, the fiber orientations and the stacking sequence. The fiber orientations are chosen from some prepared angles. In the proposed procedure, the rigidity of laminated composites is approximated by means of weighing coefficients and transformed elastic constants at the prepared angles. The proposed approximate function can be thought of response surfaces, because the weighing coefficients are adjusted with pairs of input and output data. The effective approximation is generated using auxiliary information about the principal stress distribution. The approximate function is used as the objective function of GA and can be evaluated without finite element analysis. As a result, the proposed procedure can reduce computational cost remarkably. In this paper, the detail of our developed design procedure and data sampling scheme are described. Key Words:genetic algorithm, function approximation, laminated composite, stiffness design, fiber orientation, stacking sequence, computational cost