Multi-Agent Systems for Industrial Diagnostics

Abstract Industrial diagnostic systems aim at anticipating the occurrence of failures or, should failures have occurred, at detecting them and identifying their cause. Similar to a process control system, a diagnostic system is closely connected with a specific industrial application. In case of large and complex industrial systems, such intimate link translates into the need to satisfy requirements for modularity, flexibility, and adaptability. This paper introduces a multi-agent software architecture as a possible solution to better tackle the variety of problems connected with building diagnostic systems, based at the requirements connected with modem industrial applications.