Analysis of Dynamic Robust Design Experiment with Explicit & Hidden Noise Variables

Abstract

The response model approach proposed in Tsui [25] allows greater flexibility to investigate factor effects for analyzing dynamic robust design experiments. This article generalizes the response model approach based on a generalized linear model. We develop a generalized two-step optimization procedure to substantially reduce the process variance by dampening the effect of both explicit and hidden noise variables. The proposed method provides more reliable results through iterative modeling of the residuals from the fitted response model. The method is compared with three existing approaches (the response model, the loss model, and the response function model) in two practical examples.

Key Words:

• Generalized linear models (GLMs)
• response model (RM) approach
• Taguchi method
• Two-stepoptimization

Additional information

Notes on contributors

Suk Joo Bae

Suk Joo Bae received his B.S. and M.S. degrees in the Department of Industrial Engineering at the Hanyang University, Seoul, in 1994 and 1996 respectively, and his Ph.D. in the School of Industrial and Systems Engineering at Georgia Institute of Technology in 2003. He is currently an assistant professor in the Department of Industrial Engineering at Hanyang University, Seoul, Korea. He worked as a reliability engineer at the Samsung SDI, Korea, from 1996 to 1999 and as a Post-doctoral Research Associate at the University of Tennessee, Knoxville. His research interests are centered on reliability evaluation of light displays and nano-devices via accelerated life and degradation testing, statistical robust parameter design, and process control for large-volume on-line processing data. He is a member of INFORMS, ASA, and IMS.

Kwok-Leung Tsui

Kwok-Leung Tsui is professor in the School of Industrial and Systems Engineering at Georgia Institute of Technology. He has a B.Sc. in Chemistry and an M.Ph. in Mathematics both from the Chinese University of Hong Kong, and a Ph.D. in Statistics from the University of Wisconsin at Madison. He had worked in the Quality Assurance Center of AT&T Bell Laboratories in 1986–90. Dr. Tsui was a recipient of the 1992 NSF Young Investigator Award. He is a Fellow of the American Statistical Association (ASA), and was the (elected) President and Vice President of the ASA Atlanta Chapter in 1992–1993. Dr. Tsui was the chair of the INFORMS Section in the Quality, Statistics, and Reliability (QSR) in 2000 and was the program chair of the QSR cluster sessions in 1999 and 2000. He is the founding chair of the INFORMS Section in Data Mining (DM) and was the program chair of the DM cluster sessions in 2002 and 2003. Dr. Tsui is also a US representative in the ISO Technical Committee on Statistical Methods (TC 69). Dr. Tsui researches, teaches, and consults on statistical methods for quality, logistics, and data mining. His research interest includes classification tree, support vector machine, Mahananlobis-Taguchi System, inventory forecasting and control, statistical process control, experimental design, robust design and Taguchi method, design and modeling of computer experiments, and coordinate measuring machine modeling. Dr. Tsui has consulted for a variety of local and natinonal companies, such as AT&T, Motorola, Philips, SAIC, Six Cotinents Hotels and Resorts, Manheim Corp., and Merck.