大相岭隧道断层破碎带围岩变形的GA-BP神经网络预测技术  

Prediction of Surrounding Rock Deformation of the Daxiangling Tunnel in Fault Zones Using the GA-BP Nerve Network Technique

作　　者：张志强[1] 李化云[1] 阚呈[1] 盛越[1] 孙飞[1] 

机构地区：[1]西南交通大学交通隧道工程教育部重点实验室,成都610031

出　　处：《现代隧道技术》2014年第2期83-89,共7页Modern Tunnelling Technology

基　　金：国家自然科学基金(No.51278427);高铁联合基金(U1134208);中国中铁股份有限公司重点科技开发计划重大课题(No.重大-18-2011)

摘　　要：断层破碎带围岩变形预测一直是隧道工程信息化施工与管理争论的焦点,迄今没有科学合理的预测方法。在NATM隧道施工中,围岩变形量常作为评判隧道稳定性和支护结构经济合理性的重要指标。隧道变形量是随时间而变化的数列,可以建立实时跟踪预测模型和方法。文章针对雅西高速公路大相岭隧道施工中断层破碎带围岩变形速率大、持续时间长、变形量大等特点,基于BP人工神经网络原理,并引入了遗传算法,通过改进和提高预测精度,建立了预测断层破碎带围岩变形的GA-BP神经网络综合预测模型。经工程实践检验预测结果,具有较高的可靠性和准确性。Predicting surrounding rock deformation in fault zones has been much debated in ohservation-based tunnel construction and management, but so far a scientific and rational method for this sort of prediction is lacking. For NATM tunnelling, surrounding rock deformation is often used as an important indicator to judge the stability of a tunnel and the economic rationality of tile supporting structure. As surrounding rock deformation is a kind of series varying with time, a prediction model is established to trace and predict deformation in real time. Considering the large deformation rate of the surrounding rock around the Daxiangling Tunnel on the Ya＇an- Xichang expressway, a BP artificial neural network based genetic algorithm is introduced, improving prediction accuracy by modifying the basic genetic algorithm. Using GA-BP neural network techniques, a comprehensive model to predict rock deformation in a fraetured fault zone is established and applied to the Daxiangling tunnel, and the predicted results are verified to be accurate and reliable.

关 键 词：隧道工程 断层破碎带 GA—BP神经网络 变形预测 

分 类 号：U452.1[建筑科学—桥梁与隧道工程]