摘要
由于复杂环境制约,地铁盾构邻近桥梁风险评价过程中存在大量的随机不确定性和冗余不确定性。本文结合粗糙集理论和贝叶斯网络构建地铁盾构施工邻近桥梁安全风险评估模型:利用云模型将连续型影响因素离散化以简化数据结构;基于信息熵进行属性约简以删除冗余属性;提出基于贝叶斯网络的事前风险等级预测、事中敏感因素识别、事后致因诊断方法,为制定桥梁风险控制措施提供依据。最后结合广州地铁六号线判定邻近桥梁的风险等级,根据敏感性分析辨识出覆跨比、桥桩与隧道相对水平位置等敏感致险因子,并明确风险发生的前提下最可能的致险因素为覆跨比、桥桩与隧道的相对垂直位置和施工方法。
Due to complex environmental constraint,mounts of random uncertainty and redundant uncertainty exist in the risk assessment of shield adjacent bridge. The rough set was combined with Bayesian Network to establish the bridge safety risk evaluation model. Continuous influencing factors were dispersed to simplify the data structure according to cloud model. Based on information entropy,attribute reduction was conducted to delete the redundant attribute. Bayesian Network was used to assess the risk level beforehand,identify the sensitive factors in the matter,diagnose the accident causation factors afterwards,which provides the basis for bridge-reinforcing plan. This method was applied to line no. 6 in Guangzhou subway,the risk level of adjacent bridge was determined. The cover-span ratio and the relative horizontal position between piles and tunnel were considered as sensitive risk factors through the sensitivity analysis. Under the premise of risk,cover-span ratio,the relative vertical position between piels and tunnel,and construction method were the possible risk factors.
出处
《土木工程与管理学报》
北大核心
2016年第3期9-15,29,共8页
Journal of Civil Engineering and Management
基金
国家自然科学基金(51378235
71571078)
湖北省自然科学基金(zrz2014000104)
教育部博士后基金(2015M570645)
武汉市建委科技项目(201217
201208
201334)
