【作者】 王晓睿；

【导师】 王元汉；

【作者基本信息】 华中科技大学 ， 桥梁与隧道工程， 2009， 博士

【摘要】 在新奥法（New Austrian Tunnelling Method-NATM）施工过程中对大埋深长隧道围岩开挖进行动态监测，根据监测信息较为准确地反演分析围岩物理力学参数和初始应力场，以及对围岩类别、岩体的变形和受力状态进行分析，是围岩稳定性分析的关键。智能算法在该研究领域表现出强大的功能，成为近些年岩土工程反分析以及相关研究的重点。免疫算法（Immune algorithm，简称IA）作为一种新的智能计算方法，其在岩土工程的应用尚处于摸索阶段，然而生物免疫系统和地下工程的特点是具有很大可类比性的，因此，在反演算法中引入了生物免疫系统信息处理机制，在地下工程分析中有着更为广阔的应用前景。本文立足工程实用性，在以下方面进行了研究：1、由于地质勘探单位给予设计、施工单位的都是综合围岩分类的结果，导致设计施工单位在最不利条件下的围岩来进行设计施工，造成很大的浪费。本文对两郧断裂带的特点及其对云岭隧道施工的影响进行了分析，模拟行业专家在判断设计过程的动态思考方式，形成了基于自组织竞争神经网络及模糊推理的评判系统，根据不同地质条件，快速提出围岩分类及不同的设计方案，符合现代动态设计理念，体现了新奥法的思想精髓。2、在高应力软弱围岩中构筑长大隧道，围岩大变形必然是一个突出的问题。通过合理布置监测点，对云岭隧道围岩进行监控量测，并对数据进行处理分析。考虑到监测数据误差对分析结果将造成不利的影响，将未确知数学理论引入到位移观测数据的误差分析中，使用改进的未确知滤波法对数据进行处理。3、对围岩监测数据的分析和变形预测方法进行了介绍，通过监测数据处理，对云岭隧道围岩的收敛变形特点和软弱围岩的流变特性进行分析；并通过数据建模，对围岩进行超前预报，获取收敛趋于稳定的值和趋于稳定的时间，并在实际测量中得到验证。4、对生物免疫系统的工作机理进行介绍，分析了人工免疫算法与生物免疫系统的对应关系。设计了人工免疫算法的智能计算框架及计算流程，将所求解问题特征的相关因素，相应的计算模型，设计的智能算法进行综合考虑，得到智能化的信息处理效果，并将效果进行适当评估总结。5、考虑到隧道工程的特点和目前免疫算法的研究方向，有针对性地分析研究了免疫遗传算法、自适应免疫算法设计方法以及实现方式，引入了相似性的矢量距的抗体选择机制和自适应免疫策略；把围岩破坏模式作为一个复杂的非线性系统的辨识问题进行识别分析，结合免疫网络识别自己、非己和抗体的多样性以及处理抗原的机理，神经网络优化、聚类的特点，提出了通过免疫算法对神经网络进行优化的分析模型，研究了免疫神经网络的算法设计步骤及实现方法。6、研究了围岩物理力学参数的正演反分析的实现过程，提出模块化的基于自适应免疫神经网络的算法；形成基于FLAC^3D的正向计算—参数智能反分析—反分析结果验证的智能分析方法。在算法的实现过程中研究了如何用聚类算法确定网络的初始权值集合，如何进行免疫选择以及如何构造免疫算子的方法；通过对云岭隧道Ⅲ类围岩物理力学参数分析，反演得到的围岩物理力学参数改进了初勘资料的建议值，这对隧道围岩稳定性评价、信息化设计具用实际意义。更多还原

【Abstract】 In the construction for deep and long tunnel with New Austrian Tunneling Method（NATM）,it is important to monitor dynamically the surrounding rock mass.According tothe monitor information,the physical and mechanical parameters and the initial stress fieldof rock mass can be obtained by the back analysis.Then the rock type,rock deformationand state of stress could be analyzed.This is the key step for the analysis of the stability ofthe surrounding rock mass.In the analysis the intelligent algorithm （IA）has shown itsstrong role and has become an important field of the back-analysis and related studies ofgeotechnical engineering.Immune algorithm （Immune algorithm,referred to as IA）as a new intelligent methodof calculating its application in geotechnical engineering is still in the exploratory stage,However,the biological immune system and underground works are characterized by agreat analogy can be.Therefore,the introduction of anti-algorithms biological immunesystem information-processing mechanism has a more wide application prospect inunderground engineering analysis.This dissertation based on practical projects,in thefollowing areas was studied:1.As the geological prospecting unit offers the results of a comprehensiveclassification of rock mass to design and construction,it may led to design andconstruction in the most unfavorable conditions of the surrounding rock mass and result ina great waste.In the studies,the influence of the fault zone of Yunxi and Yunxia on thecharacteristics of Yunling Tunnel construction is analyzed.Set up the judge system offuzzy inference （JSFI）based on the Self-Organizing-Feature-Map and fuzzy math to dealwith the uncertain information.The classification of the surrounding rock mass and thesupport program can be pointed out according to the geologic conditions.It coordinateswith the dynamic design of the tunnel construction and represents the soul of NATM.2.In the high geo-stress of weak rock,the large rock deformation is an obviousproblem to build the long tunnel.In the process of Yunling Tunnel construction,it needsto arrange the measuring points,monitor the measurement of the surrounding rock mass,and analyze the data processing.In the analysis of the monitoring data,the gross error ofanalysis is taken into account to the results.The unascertained mathematics is introducedinto the error analysis of the displacement measurement data,and the improved filteringmethod is used in the data processing.3.The methods of monitoring data analysis and deformation prediction of rock massare introduced.By monitoring the data processing,the convergence of the surroundingrock of the deformation characteristics and rheological behavior of the soft rock are studied.Based on the data analysis,the rock mass prediction is carried out,and theconvergence time and values are determined.The results can be verified by the practicalmeasurement.4.The biological immune system mechanism is introduced.The relationship betweenthe artificial and biological immune system are studied.The relevant factors of thecharacteristics with solving problem,the computing model and the design of intelligentalgorithms,the designed computing framework of artificial immune algorithm intelligentand calculation process are considered.The intelligent information treatment is obtainedand the effects are evaluated.5.Taking into account the characteristics of the tunnel project and the current trend ofthe immunization research,the targeted analysis of the immune genetic algorithm,theadaptive immune algorithm design as well as the way to achieve are studied.The antibodyselection mechanism of the similarity of the vector and adaptive immune mechanismsstrategy are introduced.The immune network ability of identify themselves and non-self,antibody diversity and the mechanism to deal with antigen,neural network optimizationand clustering features are combined.Through the immune algorithm for neural networkoptimization model,the immune neural network algorithm design and implementationsteps are studied.6.The intelligent direct back analysis of physical and mechanical parameters of thetunnel surrounding rock mass is studied based on the adaptive immunity algorithm and BPneural network.An intelligent analytic method based on the process of FLAC^3Ddirectcomputation-parameter intelligent direct analysis-back analysis results verification isproposed.In the process of the realization of algorithm,a series of problems,such as howto use the clustering algorithm to determine the weight of the initial collection of network,and how to choose the immune and structure the immunity operator,have been solved.Applying the intelligent analysis to the analysis of Yunling Tunnel rock mass,the resultshave been improved for the proposed values from the preliminary investigation.It showsthat the proposed methods have important role for stability evaluation of the tunnelsurrounding rock mass and information design.更多还原