A distributed-based approach for diagnosis of manufacturing systems

Complex systems as large manufacturing systems, telecommunication controls, or process control systems require distributed monitoring and diagnosis. We introduce a distributed multi-agent approach for diagnosis of manufacturing systems. In this approach, each agent can compute a local minimal diagnosis related to the component that would supervise and knowledge only of its own system description. By compiling diagnoses for groups of agents based on the distributed connectivity of agents, we synthesize the local diagnoses into a globally-sound system diagnosis using a message-passing approach between different categories of agents.