Towards a semantic-based process supervision

Abstract

This thesis attempts to exploit the deductive capabilities of the semantic rea- soners to automate the supervision task through a knowledge-driven approach. With that aim, we have explored the characteristics of DL-based modeling and reasoning to support qualitative supervision methods. The emphasis have been placed in multivariate data analysis. Through them, failures are detected and diagnosed using patterns of qualitative symptoms (i.e. Fault Signatures) that involve several process variables.