The study of robust design GLM of second-order response surface in noise factors

The purpose of this paper was to provide a method for disposing response surface with residual error heterogeneity by accommodating generalized linear modeling. The approach of Jan Engel and A. Freek Huele was taken as a starting point, treating noise variance as fixed effects. It was, however, assumed that process variability was caused by not only by the linear effects of noise factors but also by the reaction effects between noise factors. Therefore, a model of second-order response surface in noise factors is established. The model is fitted by iteratively reweighted least squares. Moreover, the paper discusses the model's statistics analysis and test of hypothesis.