A Study Of Binary Decision Diagram Variable Ordering Method Focused On Phased Mission System

Many safety-critical systems encountered in nuclear, aviation and electronic industries are actually phased mission systems, which are important parts of reliability engineering. Fault Tree Analysis (FTA) has been widely used for quantitatively and qualitatively analysis on the reliability of a system. An efficient way to analyze a fault tree is Binary Decision Diagram (BDD) method, which can take advantage of computer technology to automatically deal with a large scale fault tree. In recent years, the BDD approach has been widely used for phased mission system (PMS) reliability analysis. To successfully use BDD technique, a key issue is the variables ordering while converting the fault tree into BDD format. This will impact the finalized BDD size. Compared with the single phase system, the phased mission system is more complex due to the independency between its phases. So it may be not suitable if using the existed BDD variable ordering approaches for PMS-BDD construction. A new PMS BDD variable ordering method is proposed in this thesis. Considering the structure and characteristic of PMS, this new method emphasizes the adjacency of neighbor variables'ordering and simplifies the BDD construction process through common failure components reduction. By increasing shared BDD nodes and reducing repeated BDD nodes, the method proposed in this thesis could help to build a proper BDD structure for PMS. Compared with the existing best BDD ordering heuristics applied in single phase system, the proposed method can achieve performance improvement on 80% of test cases. A PMS example given in the end manifests the advantage of this new method.