中国机械工程

• 机械基础工程 • 上一篇    下一篇

一种新型动态贝叶斯网络分析方法

陈东宁1,2;侯安农3;姚成玉3;侯鑫1,2;邢然3   

  1. 1.燕山大学河北省重型机械流体动力传输与控制重点实验室,秦皇岛,066004
    2.先进锻压成形技术与科学教育部重点实验室(燕山大学),秦皇岛,066004
    3.燕山大学河北省工业计算机控制工程重点实验室,秦皇岛,066004
  • 出版日期:2020-06-25 发布日期:2020-07-22
  • 基金资助:
    国家自然科学基金资助项目(51975508,51675460);
    中国博士后科学基金资助项目(2017M621101)

A Novel Dynamic Bayesian Network Analysis Method

CHEN Dongning1,2;HOU Annong3;YAO Chengyu3;HOU Xin1,2;XING Ran3   

  1. 1.Hebei Provincial Key Laboratory of Heavy Machinery Fluid Power Transmission and Control,Yanshan University, Qinhuangdao, Hebei, 066004
    2.Key Laboratory of Advanced Forging & Stamping Technology and Science (Yanshan University),Ministry of Education of China, Qinhuangdao, Hebei, 066004
    3.Key Laboratory of Industrial Computer Control Engineering of Hebei Province,Yanshan University, Qinhuangdao, Hebei, 066004
  • Online:2020-06-25 Published:2020-07-22

摘要: 为充分发挥T-S动态故障树和动态贝叶斯网络分别在分析建模与推理计算方面的优势,提出了一种新型动态贝叶斯网络分析方法——基于T-S动态故障树的动态贝叶斯网络分析方法。将T-S动态故障树转化为动态贝叶斯网络有向无环图,再将T-S动态门及其描述规则转化为动态贝叶斯网络条件概率表,进而提出了正向推理叶节点失效概率、反向推理根节点后验概率和求解根节点概率重要度、关键重要度、风险业绩值、风险降低值、微分重要度与灵敏度的新型动态贝叶斯网络算法。通过与基于Dugan动态故障树的动态贝叶斯网络分析方法和静态贝叶斯网络分析方法对比,验证了所提方法的可行性。最后,用所提方法对液压缸同步系统进行可靠性分析,计算得到系统失效概率、根节点后验概率、重要度与灵敏度,为提高系统可靠性和进行故障诊断提供依据。

关键词: 动态贝叶斯网络, T-S动态故障树, 重要度, 灵敏度, 可靠性分析

Abstract: In order to give full play to the advantages of T-S dynamic fault tree and dynamic Bayesian network in analysis modeling and reasoning calculation respectively, a novel dynamic Bayesian network analysis method, namely dynamic Bayesian network analysis method, was proposed based on T-S dynamic fault tree.First, a T-S dynamic fault tree was converted into a dynamic Bayesian network directed acyclic graph and a T-S dynamic gate and the description rules were converted into a dynamic Bayesian network conditional probability table.Then, the algorithm of novel dynamic Bayesian network was proposed for forward reasoning leaf node failure probability, backward reasoning root node posterior probability and solving root node probability importance measure, criticality importance measure, risk achievement worth, risk reduction worth, differential importance measure and sensitivity.The feasibility of the proposed method was verified by comparing with dynamic Bayesian network analysis method based on Dugan dynamic fault tree and static Bayesian network analysis method.Finally, the reliability of hydraulic cylinder synchronous system was analyzed by the method proposed herein.Failure probability of the system, posterior probability, importance measures and sensitivities of root nodes were obtained, which may provide basis for improving system reliability and fault diagnosis.

Key words: dynamic Bayesian network, T-S dynamic fault tree, importance measure, sensitivity, reliability analysis

中图分类号: