Application of genetic algorithm to numerical experiment in robust parameter design for signal multi-response problem

Abstract

The conventional approach by Dr. Taguchi has been known as an off-line quality control to improve the performance of products or processes at low cost. Most previous Taguchi method applications have dealt only with static single-response problems. But in the recent years, the design has more complicated requirements; not only have more than one quality characteristic, but also deal in dynamic system. Therefore, several approaches dealing with multi-response problems have been increasingly reported. However, they have focused on the static system; the multi-response problems in dynamic system still have received only limited attention. This research proposes an approach using the genetic algorithm to find the best setting of the controllable factors, and outer orthogonal array to consider noises. The objective is to minimize the total average quality loss for the dynamic multi-response numerical experiment. In addition, to enhance the capability of the proposed approach, identifying the adjustment factor according to the two-step method by Dr. Taguchi is applied for the case of changing products or processes’ requirements in the future. The effectiveness of the proposed approach will be demonstrated by an example of a temperature control circuit.