摘要
传统故障树分析的二元决策图方法受变量排序影响很大,可能使运算量和运行空间过大。为此,提出在故障树预处理的基础上,使用直接生成法以简化二元决策图的生成过程。该方法避免了变量排序的影响,提高了复杂故障树的分析效率。计算中首先通过结构化简和线性时间算法模块化完成预处理,然后将直接生成法应用于各个模块进而应用到整个故障树。最后将此方法与模块连接法、相邻底事件优先法进行实例比较,证明其生成二元决策图的规模更小且运算量更少。
Traditional binary decision diagram ( BDD) method for fault tree ( FT) analysis is dependent on variable ordering, which causes large calculation and memory. To settle this problem, the paper proposes direct generation method on the basis of fault tree pre-progressing, which can avoid the influence of variable ordering and improve the efficiency of transforming FT to corresponding BDD. In the pre-processing, the original FT is simplified and modularized using linear-time algorithm. Then the direct generation method is firstly used in every module and then used in the whole FT. Finally, the direct generation method is compared with module connection method and neighbor-event-first method. The result shows that the direct generation meth-od obtains smaller number of the BDD nodes and costs less calculation.
出处
《人民长江》
北大核心
2014年第23期64-69,共6页
Yangtze River
基金
国家自然科学基金重点项目(51138001)
关键词
故障树
二元决策图
模块化
直接生成法
fault tree
binary decision diagram
modularization
direct generation method