Distributed diagnostics and dynamic reconfiguration using autonomous agents
Last modified: June 7, 2006
Abstract: Utility distribution systems such as for water and power are critical to the US infrastructure and are typically essential elements for manufacturing, commercial, and marine systems. Autonomous agents have been shown to provide superior performance in response to system faults through dynamic re-configuration to minimize loss or consequential damage while maintaining critical services. Autonomous agents may also collaborate and exchange information to detect faults or impending failures, classify the fault type, and isolate the failing component or sections in systems that are not completely instrumented. This paper describes the use of distributed autonomous agents to detect a leak in a fluid handling system and dynamically re-configure the system to and maintain critical system services. Fault detection and fault prediction are determined through collaborative decision-making among the distributed agents. The system does not employ a central controller and full system instrumentation is not required. Examples are provided showing the dynamic response of the system when arbitrary faults are injected into the system. Corresponding complexity and performance measures are provided. Simulation models showing distributed diagnostics and dynamic re-configuration with autonomous agents is demonstrated for a shipboard chilled water system and a municipal water distribution system.