【作者】 周小宏；

【导师】 吴庆鸣；

【作者基本信息】 武汉大学 ， 机械设计及理论， 2005， 硕士

【摘要】 随着TBM广泛应用于隧洞施工,致使基于TBM施工的隧洞超前地质预报研究日显突出。隧洞超前地质预报是进行隧洞施工、预防地质灾害和进行支护的基础。但是,由于隧洞地质的结构特征、岩体的性质很难事先准确判定,因而给隧洞施工、预防地质灾害带来一定的困难。 论文通过分析影响岩体质量的因素,研究国内外多种隧洞超前地质预报方法,综合对比各种预报方法预报和收集到的地质资料,提出了一种将超前钻探(冲击钻机和岩芯钻机)预报与岩心收集相结合,综合地质编录及中长期TSP(Tunnel Seimic Prediction)预报结果进行超前地质预报的的方法,对掌子面前方各种地质资料进行集中整理,综合分析,互相印证,最终判别掌子面前方地质状况。 还着重讨论了TBM隧洞施工超前地质预系统的人工神经网络预报模块。在充分了解掌子面前方围岩类别预报的重要性的基础上,利用人工神经网络处理非线性复杂关系的优势,将人工神经网络技术应用于掌子面前方围岩类别预报中,并对其中的输入输出参数的选择、样本收集与处理、BP网络模型的构造及其优化设计。通过对样本库的训练检验,对掌子面前方围岩类别预报取得了满意的效果。更多还原

【Abstract】 With TBM is applied in the tunnel construction comprehensively, study of advance geological prediction on the basis of TBM construction overhangs. Advance geological prediction is foundation of tunnel construction, preventing geology casualty and support. However, It is difficult to determine the geological structure character and rock masses character, which makes the precautionary geology casualty and the construction technique difficult.The paper analyses effect factors of rock qualitation and methods of advanced geological prediction abroad and home. On basis of integration and comparison them, one methed of advanced geological prediction is offered, which integrates the geological survey, TSP(Tunnel Seimic Prediction) methed, forward drilling and porbing (including percussive drilling and core drilling) and acoustic sounding are brought forward to applying to practical engineer. The method makes central arrangement, comprehensive analysis and mutual acknowledge to distinguish geologic condition forward mass.In this dissertation, the artificial neural network module of advance geological prediction system in TBM construction tunnel is also emphasized on. On the basis of fully realizing the importance of predicting the rock masses classification in front of the tunnel face, artificial neural network is used to predict the rock masses classification, and mainly discusses the preferences, stylebook collection and disposal, and how to structure BP neural network module and its optimization, and compares the effects of adding drilling speed eigenvector with not adding, thanks to BP neural network’ s superioriry in dealing with non-linear complicated relation. By training and check up of the stylebook, it’ s satisfying to predict the rock masses classification in front of the tunnel face.更多还原