基于动态故障树分析的容错系统机由测试诊断策略设计

由于大量采用各种容错技术，复杂航空航天系统具有动态随机性故障，已不再适合采用基于静态故障机理的传统故障诊断策略设计方法对其进行机内测试( BIT)设计。为此，以动态故障树模型为基础，综合考虑可靠性和测试费用等因素，提出了一种面向复杂容错系统的BIT诊断策略设计新方法。引入费用诊断重要度( CDIF)来衡量测试费用对BIT故障诊断的影响，结合独立子树、二元决策图和Markov链对动态故障树进行定性和定量分析，据此构建出基于最小割集(MCS)/最小顺序割集(MCSS)和CDIF的BIT诊断决策树。在某容错电子系统BIT设计中的应用验证了该方法的有效性与工程实用性，与基于AO“搜索和基于贪婪算法的两种诊断策略设计方法比较可知该方法的计算复杂度更小。

Built-in Test Diagnosis Strategy Design for Fault-tolerant System Based on Dynamic Fault Tree Analysis

Complex aerospace and spaceflight system has dynamic and stochastic faults because of the wide use of various fault-tolerance technologies, the conventional fault diagnosis strategy design meth?ods based on static failure mechanism is inappropriate to its built-in test (BIT) design. In order to solve this problem, a new method for designing BIT fault diagnosis strategy for fault-tolerant systems, considering reliability and testing costs, was proposed based on dynamic fault tree model. The costs and diagnostic importance factor (CDIF) was introduced to evaluate the influences of the test costs on the BIT diagnosis, and individual sub fault tree, binary decision diagram and Markov chain were inte?grated to analyze the dynamic fault tree qualitatively and quantitatively, and consequently, the BIT di?agnostic decision tree is built based on minimal cut set(MCS) /sequence (MCSS) and CDIF. The ap?plication in BIT design of a certain fault-tolerant electronic system demonstrates its engineering practi?cability and validity. The comparison of the fault diagnosis strategy design methods among the pro-posed and two other methods shows that the proposed is better than those based on AO * algorithm and greed algorithm in respects of computational complexity.