Dynamic Fault Tree Analysis Method Based On Modular Thinking

With the quick development in science and technology, system structure is becoming increasingly sophisticated. The traditional fault tree analysis methods are inapplicable to the system reliability analysis of dynamic random faults and relativity. During dynamic system reliability analysis, Markov Model often leads to NP problem. The dynamic fault tree analysis method based on modularization idea will make dynamic fault trees modularized to gain an independent static subtree and dynamic subtree respectively, then realize solving via Binary Decision Diagram (BDD) method and Markov chain method, thereby raising the analysis efficiency greatly. However, there have not been a single and universal separate module searching method, or a BDD based static subtree analysis method or a method which can realize automatic conversion from a dynamic subtree to Markov chain yet. This article focuses on the analysis and study of these problems.In terms modularization, this article introduces a linear search method for separate modules, which shortens the module searching time greatly. In terms of static subtree, this article first introduces a method with which the solving of the optimum indicator sequence of fault tree bottom event is realized on the basis of structure importance and then realizes the conversion from a static subtree to BDD by means of recursive algorithm. On this basis, this article also introduces qualitative analysis and quantitative analysis methods about the BDD based fault tree, proposing a method to solve the non-coherent fault tree via the BDD early non-intersection. To solve the problem that a dynamic subtree is hard to convert into Markov chain directly, this article proposes a method combining combinative analysis and recursive search based on timing constraint and fault constraint, thus realizing the automatic conversion from a dynamic subtree to Markov chain. Moreover, this article conducts further study on the algorithms of the qualitative analysis and quantitative analysis of dynamic subtrees based on Markov chain. Finally, this article shows us the convenience and practicability of the methods above via the analysis calculation of the actual fault tree of a system.In order to make the software implementation of dynamic fault tree analysis methods easy and make them apply to projects better, this article has finished the framework designing of the dynamic fault tree analyzing platform and focuses on the detailed description of its analysis calculation module.