Causation-based T ² decomposition for multivariate process monitoring and diagnosis

Successful application of SPC for monitoring and diagnosis in complex manufacturing systems is a challenging problem due to the excessive number of process/product variables and their intricate relationships. The T ² decomposition approach [7] (called MTY approach), although popularly used, may receive concerns on its computational efficiency and capability of reliably identifying the root causes of the fault in a large complex system. This paper proposes a causation-based T ² decomposition approach by effectively incorporating causal models into the MTY approach. Theoretical analysis and simulation studies demonstrate that the proposed causation-based T ² decomposition has enhanced interpretability and diagnostic power.