Analysis of Designed Experiments with Multichannel Profiles Response Variable

Please enable javascript in your browser.

Badfar, Mohammad | 2017

599 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 49785 (01)
University: Sharif University of Technology
Department: Industrial Engineering
Advisor(s): Niaki, Akhavan
Abstract:
The purpose of this research is analyzing designed experiments which their response variable is in form of multichannel profiles. For this purpose, a number of experiments with multichannel profile response variable designed at first. Then by random effect model, output data calculated. Experiments output data dimension reduced using principal component analysis and its extensions. After that, regression analysis used to analyze results of dimensionality reduction data in order to estimate coefficients of potentially effective variables in response. At the end, coefficients of effective variables classified with a hierarchical classification method in order to discover change and its root cause in process. In this research, experiments with multichannel profiles analyzed simultaneously and it is shown that this approach is more efficient than former approach, which was analyzing designed experiments with multichannel profiles response, without simultaneously analysis or without considering interaction.
Keywords:
Dimensionality Reduction ; Principal Component Analysis (PCA) ; Regression Analysis ; Profile Monitoring ; Designed Experiments Analysis ; Hierarchical Classification ; Multichannel Profiles

No TOC