Structural Optimization Post-Process Using Taguchi Method

抄録

It is well known that the optimization technology is utilized widely with finite-element analysis for structural design. Generally, the sizes of structure members are regarded as design variables in structural optimization where the design variables are assumed to be continuous. Since the available designs are discrete, one-step larger existing values from the optimization solution can be taken for the final design although this process is quite inefficient. In this study, the Taguchi method which has been used in the quality control area is adopted for discrete structural design to enhance the performances of the final design. It is used in the post-process of the structural optimization. An orthogonal array is constructed with discrete values around an optimal solution and the cost function is evaluated by the Taguchi method. Unconstrained and constrained problems are selected as examples emphasizing structural design problems. Excellent solutions have been obtained for the unconstrained problems and fairly good designs are achieved with some limitations in constrained problems. Currently, the solutions by the Taguchi method are analyzed in engineering sense and they are considered excellent in practical design applications. Therefore, more efforts are needed for the mathematical evaluation and wider applications.