摘要
基于邻近既有线施工风险因素复杂多变且具有不确定性,而传统风险分析方法很难处理不确定性知识的现状问题,提出一种基于贝叶斯网络的邻近既有线施工风险分析方法。结合系统安全科学理论基于事故资料统计分析建立邻近既有线施工风险致因模型,并由专家群决策方法确定风险因素清单,在此基础上构建邻近既有线施工风险BN结构模型;结合实际工程案例,利用贝叶斯双向因果推理原理预测项目邻近既有线施工风险发生类型以及不同情况下的风险发生概率,诊断风险成因机理;通过Ge NIe敏感性分析找出敏感致险因素。研究结果表明:在人、物、环和管风险因素系统中,管理因素的变化对邻近既有线施工风险水平的影响较大。模型分析结果与项目实际情况基本相符。
The risk factors of railway construction close to the existing line is complicated and uncertain.In order to solve the deficiency problem that traditional risk analysis methods have in dealing with uncertainty,a risk analysis method of railway construction adjacent to existing line was proposed.Firstly,a risk causation model was established through System safety science theory and statistical analysis of accident data in railway construction close to existing line,and the list of risk factors was determined by expert group decision making method.On this basis,the BN structure model of construction risk close to existing line was constructed.Then,combined with the actual engineering case,the types of construction risk close to existing line and probabilities under different conditions were predicted using forward causal reasoning,and formation mechanism of construction risk close to existing line was analyzed with reverse diagnoses reasoning.Finally,the sensitive risk factors were found with sensitivity analysis in GeNIe.The results show that the management factors have greater impact on the risk level of construction close to existing line compared to the other risk factors in the system of human,object,environment and management.The analysis results are basically consistent with the actual situation of the project.
作者
田苾
黄健陵
陈辉华
杨丁颖
TIAN Bi;HUANG Jianling;CHEN Huihua;YANG Dingying(School of Civil Engineering,Central South University,Changsha 410075,China)
出处
《铁道科学与工程学报》
CAS
CSCD
北大核心
2018年第8期2163-2171,共9页
Journal of Railway Science and Engineering
基金
国家自然科学基金资助项目(51378509)
关键词
风险管理
铁路施工
临近既有线
贝叶斯网络
不确定推理
敏感性分析
risk management
railway construction
close to existing line
Bayesian network(BN)
uncertain reasoning
sensitivity analysis
