| In PART I, optimization method for dynamic robust design experiment is illustrated. This article generalizes Response Model (RM) approach in the dynamic system. Based on the Generalized Linear Model (GLM), generalized two-step optimization procedure is introduced to further reduce the process variance by identifying the effects of hidden noise variables. Our suggested method (called GLM-RM approach) provides more reliable results by modeling the residuals from the fitted response model simultaneously.; In PART II, various modeling approaches are introduced to estimate failure time distributions of degradation testing data precisely. Two kinds of modeling approach; Nonlinear Random coefficients (NRC) models and a (log) linear model with a change-point are used to describe the nonmonotonic degradation path. We show that reliability estimation can be substantially improved by considering burn-in characteristics of light displays. |
