Methods based on independent component analysis for monitoring industrial processes

• Grant number: 11/23546-3
• Support Opportunities: Scholarships abroad - Research Internship - Scientific Initiation
• Start date: April 01, 2012
• End date: July 31, 2012
• Field of knowledge: Engineering - Electrical Engineering
• Principal Investigator: Leonardo Tomazeli Duarte
• Grantee: Guilherme Dean Pelegrina
• Supervisor: Christian Jutten
• Host Institution: Faculdade de Ciências Aplicadas (FCA). Universidade Estadual de Campinas (UNICAMP). Limeira , SP, Brazil
• Institution abroad: Université de Grenoble, France
• Associated to the scholarship: 11/14025-0 - On the application of independent component analysis to the problem of process monitoring through control charts, BP.IC

Abstract

Control charts are commonly employed to determine whether the variability observed in a process stems from natural causes or assignable causes, such as machine failure. When an assignable cause is detect, it is quite helpful to classify the nature of such a cause, since this information is useful to tackle the original problem. This task can be accomplished by automatic pattern recognition methods. The classification of assignable causes is a difficult task in the case of multivariate control charts, specially when there are more than one assignable cause. In order to overcome this problem, one may conduct a pre-processing stage (feature extraction), whose goal is to discriminate the assignable causes observed in the process. This problem will be tackled in the present project, whose goal is to investigate the application of independent components analysis methods and other techniques developed in the context of the blind source separation problem, to the analysis of multivariate control charts. (AU)