基于FTA-BN集成的长输管道失效分析

Failure Analysis of Long Distance Pipeline Based on FTA-BN Model

为了提高长输管道失效事故预防的准确率,准确监控长输管道腐蚀状态,提出了基于贝叶斯网络(BN)的长输管道失效风险动态评估法。首先基于影响长输管道失效风险的多源易变因素,系统地分析各因素间的因果依赖程度;然后基于事故树(FTA)理论构建事故树耦合贝叶斯网络的风险预测模型,并利用GeNie软件实现贝叶斯网络预测和诊断故障过程的双向推理,进而根据管道失效二态性和逻辑门确定性的分析,实现管道失效风险的薄弱环节及故障节点的快速识别;最后通过实例分析,验证模型的科学性。结果表明:FTA-BN模型较传统的FTA模型推算出管道失效基本事件的分布概率更准确高效。

In order to improve the accuracy of long-distance pipeline failure accident prevention and accurately monitor the corrosion state of long-distance pipeline,a dynamic evaluation method of long-distance pipeline failure risk based on Bayesian network(BN)is proposed.Firstly,based on the multi-source variable factors affecting the failure risk of long-distance pipeline,the causal dependence among the factors is systematically analyzed;Then,the risk prediction model of fault tree coupled Bayesian network is constructed based on the fault tree(FTA)theory,and the two-way reasoning of Bayesian network prediction and fault diagnosis process is realized by using GeNie software.Then,according to the analysis of pipeline failure duality and logic gate certainty,the weak links and fault nodes of pipeline failure risk are quickly identified;Finally,an example is given to verify the scientificity of the model.The results show that the distribution probability of pipeline failure basic events calculated by FTABN model is more accurate and efficient than the traditional FTA model.