81 巻 (2015) 832 号 p. 15-00437
An optimization technique using the hybrid GA for large structures in which a lot of stiffeners are attached is proposed in this paper. The object function is the mass of structure and the design variables are the number and the position of stiffeners, the shape of stiffener and the thickness of plate. These design variables make a combinational optimization problem for which genetic algorithm is effective. However, this optimization problem includes the position of stiffener and the thickness of plate which are kinds of continuous design variables. A lots of bits are needed for expressing continuous design variables by strings of genetic algorithm in order to have reasonably small increments of them, which results in a large calculation amount. In order to overcome this problem, a hybrid GA is introduced in this paper. Hybrid GA is optimization method which combines GA with another optimization technique which handles continuous design variables. The relation among the thickness of plate, the span of stiffeners attached on the plate, the deflection and the stress of plate are introduced in order to predict suitable design variables in this added optimize technique. The proposed optimization method makes it possible to optimize continuous design variables, the position of stiffeners and the thickness of plate,with small calculation amount. Several selected problems are solved to show the capacity and validity of the proposed method.