Decentralised Diagnosis of Automata Networks

Authors:	Neidig Jörg, Ruhr-Universität, Germany Lunze Jan, Ruhr-Universität, Germany
Topic:	6.4 Safeprocess
Session:	Safety and Structure Analysis for Fault Diagnosis/ Diagnosis of Hybrid and Discrete-Event Systems
Keywords:	Automata theory, diagnosis, decentralised systems, discrete-time systems

Abstract

Automata-theoretic models have been used successfully in model-based process supervision and diagnosis. From a practical viewpoint, their main drawback is their complexity, which increases fast with the size of the original discrete-event system. This complexity can be reduced by compositional modelling which results in an automata network. The reduced complexity of the network leads to a complexity reduction of the diagnostic algorithm, because the fault diagnosis can be performed in a decentralised way. The paper develops such a diagnostic method for nondeterministic and stochastic automata networks.