机构地区: 吉林大学计算机科学与技术学院
出 处: 《计算机学报》 2003年第9期1171-1176,共6页
摘 要: 对诊断问题的分解进行了研究 ,给出了基于模型诊断问题分解的判定定理 ,刻画了利用系统观测值和参量假定例化值分解诊断问题 ,提出了有条件可分解诊断问题的概念 ,进一步刻画了基于模型的诊断问题分解 ,对如何利用参量假定例化值分解诊断问题给出了最可能优先算法 ,并对该算法的正确性、完备性及复杂性进行了证明 . This paper investigates the decomposition of diagnosis problem, gives theorem discrimination to the decomposition of model based diagnosis problem, characterizes how to decompose diagnosis problem by using the observations of the system and the assuming instantiations of some variables, proposes the concept of conditioned decomposable diagnosis problem, further characterizes the decomposition of model based diagnosis problem, gives a best first algorithm to how to make best use of the assuming instantiations of some variables to decompose diagnosis problem, and proves the correctness, completeness and complexity of the algorithm. The results in the paper can provide the theoretical evidence for improving the effectiveness of diagnosing the tree like structured systems.
关 键 词: 模型诊断 树型结构 算法 可分解诊断问题 人工智能
领 域: [自动化与计算机技术] [自动化与计算机技术]