基于动态贝叶斯网络的TBM卡机风险预测

为了预测隧道机械施工时隧道掘进机（TBM）卡机风险，基于动态贝叶斯网络（BN）分析卡机概率系统. 通过专家知识以及解释结构模型确定风险因素和风险事件的因果关系，收集国内隧道机械施工地质条件实测数据，根据相关规范、研究成果和云模型云间划分法对风险指标进行区间划分，运用粗糙集分类原理对数据进行离散化，获取风险因素的原始先验概率和风险事件的条件概率. 结合软件GENIE，建立动态BN模型预测卡机风险. 结果表明：在无证据条件下，TBM卡机风险概率为8%；造成TBM卡机的关键风险因素是岩石类型、大量的地下水和断裂破碎带. TBM卡机的关键致因链为岩石类型→掌子面突泥涌沙→卡刀盘→卡机，大量的地下水→掌子面突泥涌沙→卡刀盘→卡机，高地应力→软岩大变形→卡护盾→卡机，围岩坍塌→卡护盾→卡机.

Risk prediction of TBM jamming based on dynamic Bayesian network

A probability system of jamming was analyzed based on dynamic Bayesian network(BN), in order to predict the risk of tunnel boring machine (TBM) jamming during tunnel mechanical construction. Through expert knowledge and explanation structure model to determine the causal relationship between risk factors and risk events, collecting the measured data of domestic tunnel mechanical engineering geological conditions. The risk indicators were divided according to the relevant specifications, the research results and the cloud division method of cloud model into internal division. The rough set classification principle was applied to discretization of data to obtain the original prior probability of risk factors and the conditional probability of risk events. Combined GENIE software a dynamic BN model was established to predict the risk of card machines. Results show that the risk probability of TBM jamming is 8% under the condition of no evidence. The key risk factors of TBM jamming are rock type, large amount of groundwater and fracture zone. The key cause chain of TBM clamping machine is as follows: rock type → progradation of mud and sand on palm face → clamping knife plate → clamping machine, a large amount of groundwater → progradation of mud and sand on the palm surface → clamping knife plate → clamping machine, high ground stress → large deformation of soft rock → clamping shield → clamping machine, surrounding rock collapse → clamping shield → clamping machine.