基于有向图故障树自动建树方法的规范化描述及其应用研究

Research on Formal Description of Digraph-Based Fault Tree Construction and Its Application

基于有向图的建树方法是目前故障树自动建树研究中最常用的方法之一 ,但其有效性多年来一直存在争议 ,其主要原因在于采用传统的专家定义的算子进行建树推理时 ,算子定义不规范 ,导致对系统复杂的控制结构适应能力不强 .该文利用人工智能的原理 ,对定性推理逻辑进行了必要的扩展 ,提出了建树过程的形式化描述 ,在此基础上将故障树的建树问题转化为一个约束满足问题 (CSP) ,从而可利用比较成熟的算法来解决上述问题 .针对实际问题的例证说明了这一过程 .研究表明 ,文中所提出的方法更便于计算机自动处理 ,减少建树过程中的人为失误 ,可有效地提高故障分析效率 .

As one of the most attractive approaches in fault tree automatic construction, the digraph-based fault tree construction is not effective for complex system due to the unsuitable traditional operators defined by domain experts. In this paper, a formal description of the construction process is developed by extending the multiple-valued logic and utilizing constraint logic programming principle. Thus, the fault tree construction is transformed to a constrain satisfaction problem (CSP). And the problem can be simply solved through the CSP algorithms. The results of the fault tree automatic construction for a classical example show our method can reduce inaccuracy and is more efficient and effective.