Another view of dual response surface modeling and optimization in robust parameter design

Abstract

Robust parameter design (RPD) based on the concept of building quality into a design has received much attention from researchers and practitioners for years, and a number of methodologies have been studied in the research community. There have been many attempts to integrate RPD principles with well-established statistical techniques, such as response surface methodology, in order to model the response directly as a function of control factors. In this paper, we reinvestigate the dual response approach based on quadratic models Vinning and Myers (J Qual Technol 22:38–45), which is often referred to in the RPD literature and demonstrate that higher-order polynomial models may be more effective in finding better RPD solutions than the commonly-used quadratic model. We also propose optimization models for each of the three classes of quality characteristics (i.e., nominal-the-best, larger-the-better, and smaller-the-better). The optimal solutions obtained using the proposed models are compared with the solutions obtained using the RPD techniques in the current literature.