Application of multivariate statistical techniques to predict process quality

Abstract

Almost all of the industrial processes are automated to collect large volumes of data that are produced during the manufacturing phase, and it is stored in the database as a backup. But, now the current trend is to understand the archived data and to obtain important information that is hidden among vast amounts of data. Such useful information obtained will further provide a detailed and better process knowledge that helps in improving the quality of the end product as per the standard specifications, and thereby, an increase in the profit can be expected. This paper focuses on two multivariate statistical techniques, principal component analysis (PCA) and partial least square (PLS), to predict the quality of an industrial process. A predictive model is built by extracting the important predictors that has a greater influence on the response quality. A case study is considered to evaluate the performance of the proposed techniques with respect to the specified quality and found that the parameters that influence the quality of product can be analyzed and predicted to better accuracy.