Application of Genetic Algorithm to Numerical Experiment in Static Robust Parameter Design

抄録

This paper discusses how to apply the genetic algorithm method with robust parameter design for determining controllable factors effectively. It adopts the traditional concepts of orthogonal array experiments that take the noise factors into account for the design stage, signal-to-noise ratio for evaluating the performance and a two-step method for determining the best condition. Even though, there are a lot of studies about genetic algorithm for robust design, but there are not enough studies on suitable performance measures and how the important two-step method applies with this approach effectively. Therefore, while taking advantage of the unique features of a genetic algorithm enhancing suitable performance measures, and incorporating a two-step method, a more effective robust parameter design is achieved. Its application is demonstrated by the use of an example "Pen recorder" to show its effectiveness. The results suggest that the genetic algorithm method with a quality loss function and the applied two-step method is an efficient approach that can aid robust parameter design in terms of time, simplification and optimum solution.