针对开放环境中多Agent合作诊断问题,本文提出了一种新的、基于TMS的冲突消解策略。在文中描述的信念知识表示基础上,研究了利用异构的Agent对同一对象进行诊断,以及诊断结果间冲突的类型和信念修正方法。该系统不同于简单的投票机制,既可以纠正错误的假设,提高诊断结果的可靠性,又可以实现知识库的一致性维护。同时也作了形式化的分析和描述。 Aiming at the problem of multi-agent cooperative diagnosis in open environment, this paper proposes a new TMS-based conflict resolution strategy. Based on the knowledge representation of belief described in this paper, the methods to diagnose the same object by using heterogeneous Agent and the types of conflict between the diagnosis results and belief correction are studied. Unlike the simple voting mechanism, this system not only can correct false assumptions, improve the reliability of diagnosis results, but also can maintain consistency of knowledge base. At the same time also made a formal analysis and description.