Automatic Partitioning of DES Models for Distributed Fault Diagnosis Purposes

The automatic partitioning of automata models for large concurrent Discrete Event Systems (DES) is considered. Distributed model architectures are widely used for automatabased fault diagnosis to ensure model completeness and to avoid the state space explosion problem. Partitioning determines the size and structure of partial models to represent concurrent operating parts of the DES. In this paper, new automatic and data-based partitioning methods are proposed using observations of the fault-free system behavior. The application on a real-world benchmark system shows the capabilities of the developed methods.