Derivation and Evaluation of Plausible Counter Actions by Combining Qualitative Reasoning, Empirical Knowledge and Numerical Simulation

摘要

This paper describes a technique to derive and evaluate plausible counter actions in an anomalous situation of engineering systems. The technique has three steps composed of a derivation step of plausible counter actions applying the qualitative reasoning based on a functional model, selection step of derived counter actions by empirical knowledge of their operabiliry, regularity, and effectivity, and evaluation step by a static plant simulator. By combining the qualitative reasoning, empirical knowledge, and numerical simulation, effective and reasonable counter actions are obtained. The technique is implemented in an operator support agent called semantic information presentation agent (SIPA). The SIPA collaborates with other agents of interface agent, virtual plant agent, large screen agent, and ontology server on a distributed collaboration infrastructure. Other tasks of the SIPA is to display necessary time-response graphs of plant signals to understand plant situation, to identify an anomaly causes and to display diagnostic information from functional viewpoint as well as from behavioral, operational, and structural ones. The applicability of the derivation and evaluation technique is discussed through several applications to a main part of an oil refinery plant.