摘要
基于模型的诊断是一项新型的智能推理技术,是人工智能领域中一个炙手可热的研究分支。提出元件输出标志的概念,通过在系统中传播输出标志,来判断元件集合是否为系统的诊断。使用SE-Tree(set enumeration tree)形式化地描述整个计算过程,逐步生成当前系统对应的所有极小诊断。此方法不求解冲突集和碰集,直接求出所有的极小诊断,与Reiter的模型诊断方法有着本质的不同,极大地减小了诊断求解的复杂度。实验结果表明,该算法具有较好的效率,并且适用于复杂的对象的诊断问题。
Model-based diagnosis is one of the active branches in artificial intelligence, which is a type of intelligent reasoning technology and it overcomes the shortcomings of traditional fault-diagnosis methods. This paper presents the concept of component output flag, then whether a component set is a diagnosis in the system or not can be driven through the propagation of the output flag. The method can compute all the minimal diagnoses directly without computing all the conflict sets and the hitting sets of the collection of the corresponding conflict sets like classical methods do. And then the combinatorial explosion caused by calling ATMS （assumption-based truth maintenance system） , known as an NP-complete problem, can be avoided as well. The computing procedure is formalized by introducing SE-Tree （set enumeration tree） to produce all the solutions gradually. Results show that the diagnosis efficiency is highly improved with this method, which satisfies real-time requirement, even for a complex system.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2011年第12期2857-2862,共6页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(60973089
60873148
60773097
61003101)
吉林省科技发展计划项目基金(20101501
20100185
20090108
20080107
201101039)
国家教育部博士点专项基金(2010006111003)
浙江省自然科学基金(Y1100191)
浙江师范大学计算机软件与理论重中之重学科开放基金资助项目
