Abstract: Fault diagnosis of case-based reasoning (CBR) as a novel intelligent technology can simulate the behavior of human in solving problems to find out the resolution for new ones. This paper outlines the theoretical basis and basic principles of CBR. Further, it surveys recent research for fault diagnosis of CBR from two aspects as working process and integrated framework, the former includes representation, retrieval and reuse of cases, as well as maintenance of case-base, and the latter has CBR with rule-based reasoning, artificial neural network or multi-agent system. Finally, the paper gives the characteristics and applications for six kinds of methods to fault diagnosis, and points the future research fields.

Key words: artificial intelligence, case-based reasoning (CBR), fault diagnosis, multi-agent

摘要： 基于案例推理(case-based reasoning, CBR)的故障诊断作为一种新的智能诊断技术, 模拟人类求解问题的思路, 通过历史案例发现新问题的解。概述了CBR 的理论基础和基本原理; 从工作过程和集成框架两个方面综述了CBR 故障诊断技术的研究现状, 其中工作过程包括案例的表示、检索和重用, 以及案例库的维护, 集成框架包括CBR 与基于规则推理、CBR 与人工神经网络以及CBR 与多智能体等三种情况; 比较了六种故障诊断技术的特点及应用范围; 总结了CBR 故障诊断技术有待解决的问题。

关键词: 人工智能, 基于案例的推理, 故障诊断, 多智能体