Monitoring and diagnosis system based in agents

This work, using the concept of agents, presents a generic model for a real time monitoring and diagnosing system and defines the agents necessary for the implementation of the monitoring and diagnosis of the system. The agents have a knowledge base that is configured by an application expert with the aid of a knowledge editor. The knowledge base implementation uses a simple, and efficient, knowledge representation structure and an inference engine able to support the monitoring and diagnosing tasks. To demonstrate the use of the model, we constructed and tested prototypes of agents to implement the fault analysis function of the Criteria Computer of the Angra 2 nuclear power plant. The results showed the viability of the use of the proposed model, indicating its potential use as a base of future nuclear power plants advisory systems. (author)