IPP-PNN模型在川藏铁路深埋长大隧道岩爆预测中的应用  

Application of IPP-PNN model in rockburst prediction occurring deep-buried long tunnel of Sichuan-Tibet Railway

作　　者：靳春玲[1] 党丹丹 贡力[1] 祁英弟 贾治元 JIN Chunling;DANG Dandan;GONG Li;QI Yingdi;JIA Zhiyuan(School of Civil Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China;Xi’an Construction Engineering Green Construction Group Co.,Ltd.,Xi’an 710000,China)

机构地区：[1]兰州交通大学土木工程学院,甘肃兰州730070 [2]西安建工绿色建筑集团有限公司,陕西西安710000

出　　处：《铁道科学与工程学报》2023年第3期986-995,共10页Journal of Railway Science and Engineering

基　　金：国家自然科学基金资助项目(51969011);甘肃省科技计划资助项目(20JR10RA274,20JR2RA002)。

摘　　要：为了准确预测在高地应力、高地温铁路隧道中的岩爆灾害,以川藏铁路前期拉林段的重要隧道节点工程为研究背景,系统、全面地总结应力水平、埋深、温度、围岩岩性及地质构造、岩体系统刚度等影响因素对川藏铁路深埋长大隧道岩爆的孕育作用,重点分析高地应力和高地温对岩爆发生的影响相关性。构建川藏铁路深埋长大隧道岩爆预测指标体系,测试并量化岩体岩爆的倾向性指标。由于各影响因素与岩爆的非线性关系,选用能充分提取数据信息、处理多因素复杂非线性问题的改进投影寻踪(Improved Projection Pursuit,IPP)评价模型对川藏铁路拉林段典型高地应力、高地温深埋长大隧道桑珠岭隧道在施工期发生的岩爆问题做初步评价,并引入密度函数估计和贝叶斯最小风险准则,将IPP模型和概率神经网络(Probabilistic Neural Networks,PNN)模型相结合,实现对岩爆等级的聚类划分。研究结果表明:根据岩爆等级预测结果可知IPP-PNN模型预测结果相比于传统PP-PNN模型和GSA-PP模型其准确度更高,在对桑珠岭隧道11~19号隧道路段的岩爆预测中,岩爆预测等级与实测等级相符合程度由66.67%和77.78%提高到100%。研究结果具有一定的应用价值和工程意义,为目前在建的川藏铁路类似隧道工程的岩爆预测提供参考。To predict the rock burst disaster in railway tunnel with high ground stress and high ground temperature,based on systematical summarization of the factors including stress level,buried depth,temperature,properties of surrounding rocks,geological structure and rock stiffness that influencing the inoculation of rock burst,the research discussed the correlation between the occurrence of rock burst and high geo-stress geothermal of the deep-seated long tunnels in Lahsa-Nyingchi section of Sichuan-Tibet Railway.Then an index system intended to predict rock burst was established,and the tendency index was tested on the basis of quantitative analysis.To tackle the non-linear relationship between the factors and occurrence of rock burst,IPP Model(Improved Projection Pursuit) which can fully extract data information and deal with complex nonlinear issues was adopted to make a preliminary evaluation of the rock burst caused by typical high geo-stress and geothermal environment in Sang Zhuling Tunnel during construction.The density function estimation and Bayesian minimum risk criterion were introduced,and the IPP model and the PNN Model(Probabilistic Neural Networks) were combined to achieve the classification of rock burst grades.Analysis result shows,according to the grades of rock burst,the accuracy of the predicted grades produced by IPP-PNN Model surpasses the traditional PP-PNN Model and GSA-PP Model.In the prediction of rock burst in the section of tunnel 11~19 of Sang Zhuling Tunnel,the conformity of the prediction with actual measurement increased to 100% from the previous 66.67% and 77.78%.The study has certain application value and engineering significance,and can provide reference to rock burst prediction in similar tunnels of Sichuan-Tibet railway under construction.

关 键 词：川藏铁路 深埋长大隧道 岩爆预测 改进投影寻踪模型 概率神经网络 

分 类 号：U459.1[建筑科学—桥梁与隧道工程]