Enhancing performance of cooperating agents in real-time diagnostic systems
Ursula M. Schwuttke, A. G. Quan · International Joint Conference on Artificial Intelligence · 1993
We present a data-driven protocol and a supporting architecture for communication among cooperating intelligent agents in real-time diagnostic systems. The system architecture and the exchange of information among agents are based on simplicity of agents, hierarchical organization of agents, and modular non-overlapping division of the problem domain. These features combine to enable efficient diagnosis of complex system failures in real-time environments with high data volumes and moderate failure rates. Preliminary results of the real-world application of this work to the monitoring and diagnosis of complex systems are discussed in the context of NASA's interplanetary mission operations.