Supervisory system for hybrid productive systems based on Bayesian networks and OO-DPT Nets

This work presents a supervisory fault tolerant system for hybrid productive systems based on Bayesian networks and OO-DPT (Object-Oriented Differential Predicate-Transition Nets) formalism. Here, systems are characterized as hybrid when they contain continuous and discrete variables and both are fundamental to describe the systems’ dynamics. The present approach introduce a procedure for the construction of models for analysis of both normal an abnormal behavior of productive hybrid systems and that can be converted in a control specification proper for practical implementation. As an example, we consider an acid recovery process of a pickle line in a steel manufacturing company. Special emphasis is laid on methodological issues and industrial applications.