Jntegration of Data Mining Algorithms and Control Charts . for Multivariate and Auto Correlated Processes

INTEGRATION OF DATA MJNING ALGORITHMS AND CONTROL' CBARI'S FOR MULTIVARIATE AND AUTOCORRELATED PROCESSES WEERAWAT JITPITAKLERT, Ph.D. ,The University of Texas at Arlington, 2009 Supervising Professor: Seoung Bum Kim The objective of tllli3 dissertation is to integrate state-of-the-art data mining 3lgoritbms with statistical process control (SPC) tools to a.chieve efficient 'monitoring in multivariate and autocorrela.ted process. Process monitoring and diagnosis have been widely recognized as important and critical tools in' system monitoring for de­ tection of abnormal behavior and quality improvement. Although traditional SPC tools are effective in simple manufacturing processes that generate a small volume of independent data, these tools are not capa.ble of handling the l.arge streams of multi ­ variate and autocorrelated data found in modern manufacturing/service systems. As the limitations of SPC methodology become increasingly obvious in the face of ever more complex processes, data mjnjng algorithms, because of their proven capa.bilities to effectively analyze and manage large amounts of data, have the potential to resolve the challenging problems that are stretching SPC to its limits. This dissertation con­ sists of two main components; data mining model-based control charts and one-class ciassilication-based control charts. First, we propose a new control chart technique that integrates state-of-the­ art data mining algorithms with SPC techniques to achieve efficient monitoring in multivari~te and autocorrelated processes. The data mining algorithms include arti-