A distributed approach for multiple model diagnosis of physical systems

Abstract

In the 1990s Agent-Based Systems (ABS) have acquired the same importance as Knowledge-Based Systems (KBS) during the 1980s. Agent technology appeal is due to the benefits that agent-based organization and coordination allow in designing cooperative problem-solving systems. In recent works on distributed artificial intelligence (DAI), researchers have tried to define more precisely the role of an intelligent agent, i.e., of the computational entity that shares the accomplishment of a specific global goal via collaborative schemes. Although important progress has been made in this field, the systematic design of DAI-oriented systems still remains a hard task. In this paper, we attempt to define a general diagnostic engine designed to allow cooperative problem solving within a web of knowledge-based agents. A framework of a cooperative diagnostic engine is proposed, where the diagnosis of physical systems is accomplished in a parallel and distributed universe of intelligent actors. To reach this goal, we formalize an organizational diagnostic knowledge structure which defines different deep models and tasks in order to better distinguish between structure and behavior. Such organization induces different cooperation schemes within the agents in order to improve the diagnostic ability in the distributed processing.