深埋高地应力隧道勘察期岩爆烈度概率分级预测

岩爆是地下工程开挖过程中硬质岩体存储的弹性应变能突然、迅速释放的动态过程。我国西南山区正在建设或拟建大量深埋长大隧道，勘察阶段岩爆的准确预测对有效设计和控制投资十分重要。从隧道工程勘察阶段线路比选与设计需求出发，针对隧道勘查期岩爆灾害预测指标获取难、预测精度低的问题，以该阶段岩爆预测指标的易获取性为前提，利用贝叶斯网络解决不确定性问题的有效性来反映岩爆烈度与各影响因素的相关关系。基于473组岩爆灾害案例，采用4个预测指标（地应力、地质构造、围岩级别和岩石强度）来构建岩爆烈度朴素贝叶斯概率分级预测模型，利用十折交叉验证方法确定模型预测精度达84.47%。将该模型应用于雅安—叶城高速公路跑马山1号隧道岩爆段落，预测结果显示：28次岩爆预测中有24次正确、4次错误，准确率高达85.71%；其中2组错误预测中，现场判别为轻微-中等岩爆，而本文模型预测为轻微岩爆。验证结果表明所建立的贝叶斯网络模型具有良好的预测性能，研究成果可为我国西南山区深埋长大硬岩隧道勘察设计期岩爆灾害预测提供技术支撑。

Probabilistic classification prediction of rockburst intensity in a deep buried high geo-stress rock tunnel during engineering investigation

Rockburst is a dynamic process of a sudden and rapid release of elastic strain energy stored in hard rock mass during underground excavation. The occurrence of rockburst disaster during tunnel construction will cause serious consequences such as casualties, equipment damage and construction delay. With a large number of deep-buried long tunnels to be constructed in southwestern mountainous areas of China, the prediction of rockburst disaster is of great importance. In this paper, to fulfil the requirement of tunnel alignment and design during engineering investigation stage, on the premise of the availability of rockburst prediction indexes in this stage, the Bayesian network is used to reflect the relationship between rockburst intensity and various influencing factors. Based on 473 groups of rockburst cases, the naive Bayesian probability classification model is constructed to predict the rockburst intensity by using four prediction indexes—geo-stress, geological structure, surrounding rock grade and surrounding rock strength. The prediction accuracy of the model is found to be 84.47% using the 10-fold cross validation method. At the same time, this model is applied to the rockburst section of Paomashan No. 1 Tunnel of Ya’an—Yecheng Expressway. The results show that the prediction accuracy is 85.71% in the 28 tunnel section applications, and the established Bayesian network model has a good prediction performance. The proposed method can provide a good support to the rockburst prediction during the investigation of deep-buried long tunnels located in Southwest China.