Causation-Based T-Squared Decomposition for Multivariate Process Monitoring and Diagnosis

• Autores: Jing Li, Jionghua Jin, Jianjun Shi
• Localización: Journal of quality technology: A quarterly journal of methods applications and related topics, ISSN 0022-4065, Vol. 40, Nº. 1, 2008, págs. 46-58
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • The Hotelling T² control chart widely used in multivariate process monitoring can effectively detect a change in a system, but cannot diagnose the root causes of the change. The MTY approach improves diagnosability by decomposing the T² statistic, but is computationally intensive and has limited capability in root-cause diagnosis when the dimension of variables is high. A causation-based T² decomposition method is proposed that integrates the causal relationships revealed by a Bayesian network with the MTY approach. Simulation studies reveal that the proposed method reduces the computational complexity and enhances the diagnosability when compared to the MTY approach.