Fault Diagnosis Expert System For Complicated Weapon System Based On Multi-Agent Technology

Agent theory is a hot topic in artificial intelligence(AI) areas. As an independent unit, agent has the following capabilities: autonomy, proactiveness, reactivity, sociality and reasoning. Such recognition helps to advance AI theory. Its combination with object oriented technology, distributed computation, artificial intelligence and other scientific areas supports intelligent system to break through the architecture of traditional expert system and localization of diagnosis theoretically and technologically. With the development of network and telecommunication, network-based distribted computation is becomming the mainstream of computer techinology.For the shortage of the traditionary centralized fault diagnosis expert system and the new features of the complicated weapon's fault diagnosis, this dissertation applies multi-agent system (MAS) theory and technology to complicated weapon's fault diagnosis and puts forword the design of fault diagnosis expert system for weapons based on MAS. It demonstrates the key technologies from sereral aspects: architecture, agent model, task decomposition and distribution, coordination and cooperation, interaction, solution synthesis. A prototype system—DFDES with opening architecture, reusable codes and modularization has been built. The system uses COM/DCOM component technology to realize communication interaction. Fault diagnosis functions rely on the cooperative performances of distributed management agents and diagnosis agents. The distribution and organization of the agents are hierarchical and multilayer structure.