제한조건이 있는 문제의 이산설계공간에서의 강건최적설계

Abstract

Engineering optimization technology has been exploited very well as an automatic design tool for the design of structures. Generally, structural optimization is carried out in the continuous space of design variables. However, the existing design variables can be discrete in the practical design. Recently, various methods for discrete design are developed including integer programming algorithms. Application is not easy since these methods are too expensive in computational cost. In this research, a relatively simple method is suggested to select discrete and robust values. The parameter design scheme of the Taguchi method is used to obtain the discrete and robust solution. The characteristic function is defined, which consists of the standard deviation and the penalty terms from the constraint violation. The orthogonal array is established around the conventional optimum and the characteristic function is evaluated. Optimum solution is selected according to the Taguchi method. The discrete optimum evaluated from these approaches has the insensitive performance to the tolerances of the design variables within the selected discrete values. Several standard structural problems are solved to show the usefulness of the suggested methods.