Possible Root Cause Finding for the Quality of Multistage Manufacturing with Principal Surface Model

Abstract

Profile data are used to assess the quality of the product in many advance process control Chang (2014) modified functional regression model and functional principal component to make functional regression model for the temporal action of two factors of profile data This interaction model explains how the quality of produce is influenced by two tool parameters in different seconds This thesis follows the method of Chang (2014) and use functional regression model and functional principal component analysis The different is that we take the time as a factor and map the data of all factors to one dimension principal curve Then we use the parameter of the principal curve to represent the influence of synchronized action from different factors at same time We use the synchronized action model to explain how the response variable was influenced by many factors Therefore the main contribution of this thesis is we use this synchronized action model to assess the influence from more than two variables at the same time for the quality of the product This thesis will introduce the functional regression model and functional principal component analysis by Chang (2014) first then illustrate how to build the synchronized action model Next we use the synchronized action model to analysis the data from Chang (2014) and illustrate the difference between the analysis result of the synchronized action model and the interaction model

Date of Award	2016 Jul 27
Original language	English
Supervisor	Shuen-Lin Jeng (Supervisor)