Methods And Implementation Of Dynamic Fault Tree Analysis

Traditional fault tree analysis method is an effective method to analyze system reliability that is based on static logic and static failure mechanism. But it cannot be used to analyze the reliability of systems with dynamic random and dependency. Markov model is often used to analyze reliability of dynamic system. However, because the size of state space will increase exponentially as the size of system augments, the construction and solution of Markov model are tedious and error prone, even cannot be used.Dynamic fault tree combines the advantages of both fault tree and Markov model. Dynamic fault tree is modularized into independent static sub trees and independent dynamic sub trees, then solved by BDD and Markov model separately. This method can improve analysis efficiency greatly if the size of each dynamic sub tree is far smaller relative to that of system.This thesis focuses on techniques of dynamic fault tree in system reliability modeling and its qualitative and quantitative analysis. It studies BDD solution for static sub trees,Markov chain solution for dynamic sub tree briefly and the modularization of dynamic fault tree;presents the algorithm for top event occurrence rate of dynamic fault tree based on Weibull distribution. Then this thesis presents a new approach to solve top event occurrence rate and a new generation algorithm of minimal cut sequence of dynamic fault tree that deviate from Markov model completely.Based on research of theory and methods, this thesis presents design scheme of the dynamic fault tree analysis platform DFTA and implement the platform; practices a dynamic fault tree analysis example of Boeing 707 plane engine oil indication system and alarm system on the platform and gets top event occurrence rate,minimal cut sequence,probability importance sequence,structure importance sequence and critical importance sequence and presents system improvement suggestion based on the analysis results.