Diagnosis of Active Systems by Automata-Based Reasoning Techniques

Abstract

This paper presents a method for the diagnosis of active systems, these being a class of distributed asynchronous discrete-event systems, such as digital networks, communication networks, and power transmission protection systems. Formally, an active system is viewed as a network of communicating automata, where each automaton describes the behavior of a system component. The diagnostic method encompasses four steps, namely system modeling, reconstruction planning, behavior reconstruction, and diagnosis generation. System modeling formally defines the structure and behavior of system components, as well as the topology of the active system. Based on optimization criteria, reconstruction planning breaks down the problem of system behavior reconstruction into a hierarchical decomposition. Behavior reconstruction yields an intensional representation of all the dynamic behaviors that are consistent with the available system observation. Eventually, diagnosis generation extracts diagnostic information from the reconstructed behaviors. The diagnostic method is applied to a case study in the power transmission network domain. Unlike other proposals, our approach both deals with asynchronous events and does not require any global diagnoser to be built off-line. The method, which is substantiated by an ongoing implementation, is scalable, incremental, and amenable to parallelism, so that real size problems can be handled.