摘要
本文给出了基于一致性的最小正常诊断的概念,并将它与基于一致性的最小反常诊断加以区别.证明了老将系统描述限制在故障理论或在故障理论中扩展有限的正常行为知识,那么使用最小正常诊断能够刻画基于一致性的诊断空间,即最小正常诊断假设成立.本文还指出:使用最小正常诊断可以缩小只针对正常行为模型的最小反常诊断所产生的诊断空间,帮助我们找到真正的故障.文中最后还指出了最小正常诊断所适合的诊断任务及诊断领域.
This paper proposes the concept of consistency-based minimal normal diagnosis and distinguishes it with consistency-based minimal abnormal diagnosis. It is shown that by restricting the description of the system to fault theory or fault theory augmented with limited knowledge of normal behavior, minimal normal diagnosis can characterize the space of consistency-based diagnoses, i. e., minimal normal diagnosis hypothesis holds. Using minimal normal diagnosis may reduce the diagnostic space produced by minimal abnormal diagnosis for models of normal behavior and can help to find the true faults. Finally, the paper points out the diagnostic tasks and application domains which are well-suited to minimal normal diagnosls.
出处
《计算机学报》
EI
CSCD
北大核心
1998年第6期560-565,共6页
Chinese Journal of Computers
基金
国家自然科学基金
国家教委博士点基金