Semester

Spring

Date of Graduation

2011

Document Type

Thesis

Degree Type

MS

College

Statler College of Engineering and Mineral Resources

Department

Industrial and Managements Systems Engineering

Committee Chair

Feng Yang.

Abstract

To eliminate unimportant factors so that the remaining important factors can be further studied in later experimentation, screening experiments (which may be physical or simulation based) are typically performed. This thesis proposes a hybrid statistical procedure for efficient factor screening via simulation experiments. The hybrid procedure is particularly developed for cases where the system response is binary, as opposed to continuous; such a factor-screening procedure does not exist yet in the literature.;The proposed hybrid procedure integrates two screening methods: the sequential factorial design with multivariate sequential test (SFD-MT), which is newly developed in this work, and the modified controlled sequential bifurcation (CSB), which is adapted from the existing CSB method. At the beginning of the procedure, a pre-screening process is conducted to obtain the preliminary estimates of factor effects, and to determine whether SFD-MT or modified CSB will be used for factor screening. Then the selected screening method (either SFD-MT or modified CSB) are performed to identify the important factors based on simulation experiments. In both SFD-MT and CSB, the type I and type II errors are approximately controlled through appropriate hypothesis tests. The efficiency of the hybrid procedure over the CSB in the literature is demonstrated via empirical experiments.

Share

COinS