This paper illustrates an application of the Disciple knowledge acquisition methodology to build an intelligent adaptive agent for monitoring the electrical power system of an orbital satellite. This methodology is used by an expert to build and train an agent in much the same way that the expert would teach a human apprentice by giving the agent specific examples of problems and solutions, explanations of these solutions, and supervising the agent as it solves new problems. During these interactions, the agent acquires general rules and concepts, continuously extending and improving its knowledge base. The agent learns by synergistically integrating the basic learning strategies: explanation-based learning, learning by analogy, and empirical inductive learning from examples. An important feature of Disciple-based agents is their ability to reason with incomplete and even partially incorrect information. The agent distinguishes between routine problems (problems it knows that it can solve correctly), innovative problems (problems to which it can recommend solutions, but for which it is not certain of the result), and creative problems (problems it is not able to solve). This allows the agent to solve a problem independently (for a routine problem), to ask confirmation of its solution (for an innovative problem), or simply to ask the expert to solve the problem (for a creative problem).
Teaching an automated agent to monitor the electrical power system of an orbital satellite
Das Trainieren eines automatischen Agenten zum Beaufsichtigen eines elektrischen Versorgungssystems für einen Orbitalsatelliten
Telematics and Informatics ; 12 , 3/4 ; 229-245
1995
17 Seiten, 7 Bilder, 2 Tabellen, 24 Quellen
Article (Journal)
English
Building an Adaptive Agent to Monitor and Repair the Electrical Power System of an Orbital Satellite
British Library Conference Proceedings | 1995
|Orbital Debris Environment Monitor (ODEM)
NTIS | 1992
|Orbital Debris Environment Monitor (ODEM)
NTRS | 1992
|Springer Verlag | 2021
|NTRS | 1966
|