1993 年 59 巻 562 号 p. 1568-1573
In this paper, an optimization technique for layout of truss structures using the genetic algorithm (GA) is proposed. It is difficult to obtain the global optimum for layout problems using the traditional gradient-based algorithms, because of the discontinuity and nonconvexity of the design space in these problems. By contrast, the genetic algorithm based on the principle of genetics and natural selection is efficient for such problems, since it makes it possible to obtain the global optimum and comparable near optimums simultaneously without using function gradients. We proposed a hybrid method combining the genetic algorithm and the generalized optimality criteria method in order to optimize the layout and cross-sectional area of truss members simultaneously. The hybrid method is adopted in the minimum weight design of the truss structures subjected to displacement constraints in which the optimum combination of truss members should be determined. From the result of numerical examples, the efficiency and reliability of the hybrid method using the genetic algorithm are discussed in comparison with the member elimination strategy starting from the ground structure.