【作者】 于国新；

【导师】 许兆义；

【作者基本信息】 北京交通大学 ， 道路与铁道工程， 2009， 博士

【摘要】 西安安康铁路秦岭特长隧道区域地质环境条件复杂,岩体结构面空间分布规律的评价与预测是指导工程设计和保障施工安全的重要基础因素。铁路隧道围岩基本分级主要由岩石的硬度和岩体的破碎程度决定,而岩体的破碎程度与岩体中结构面的分布状态和发育程度直接相关。岩体中的结构面是复杂地质构造运动的产物,它的分布具有典型的非线性特点。论文应用非线性系统科学中的混沌理论,研究了秦岭隧道区结构面空间分布非线性规律,建立了结构面预测的混沌模型,进一步从结构面分布的混沌特征入手,应用数据挖掘方法预测了秦岭隧道的围岩分级。(1)应用时间序列概念及分析方法,建立了结构面序列。为满足时间序列等间距采样要求,将结构面在测线上的间距逐一排列形成一个序列,依次可建立结构面倾向和倾角序列。由这三个序列共同确定结构面在空间的位置、倾向和倾角。用混沌等非线性研究方法,研究了结构面序列特征及其分形结构。(2)以实测数据为基础,应用非线性科学的混沌理论,研究了结构面的空间分布特征。功率谱图中低频部分负幂现象的存在、结构面发育与断层关系负幂存在、结构面分布的分形现象、Kolmogorov熵大于0等特征都证明了结构面混沌分布的存在性,Lyapunov指数趋于0、主分量分析图中先斜后趋平、功率谱负幂出现与消失又说明了结构面分布不完全是混沌,而是位于混沌的边缘。(3)相空间重构揭示结构面空间分布非线性规律,根据混沌时间序列相空间重构原理,用C-C法、预测误差最小法计算嵌入维和空间延迟量,可张显结构面在岩体中的分布特征。明确了嵌入维实际上是岩体中主要相似结构面发育的平均组数,而空间延迟是主要相似结构面之间的平均间隔等混沌学指标的物理意义。(4)通过结构面混沌特征量的计算,建立了混沌学预测模型,对结构面位置及产状进行了混沌预测,结构面序列化提供了预测的基础,相空间重构决定了混沌预测的方法。一阶局域预测法算法简单,使用方便,其预测精度不但取决于结构面序列嵌入维和空间延迟,而且与最临近相点、训练样本大小和预测步长等有关。(5)基于结构面混沌特征,进行了围岩分级预测,用已知围岩分级岩体中的结构面作为训练样本,预测出的结构面作为预测样本,以结构面切割权密度、结构面间距、倾向和倾角四个属性为预测因子,按数据挖掘理论离散化结构面数据,以基本贝叶斯分类法预测未知样本的围岩分级,工程实践结果证明是可行的。论文以非线性系统科学的混沌理论为理论基础,以知识挖掘和基本贝叶斯分类法为手段,预测了西安安康铁路秦岭特长隧道的围岩分级,研究成果不仅为该工程的设计施工提供了技术支撑,而且为其他工程中该类问题的解决探索了一个新的途径。更多还原

【Abstract】 The guiding engineering for the regional geology,complicated environmental situation,rock mass structural planes,and space pattern evaluation and surveying of the Qinling Extra-Long Tunnel of the Xi’an-Ankang Railroad is an important factor in design and construction safety.The basic classification of the surrounding rock of a railroad is determined by rock hardness and the crushing degree of rock mass,which is directly related to the distribution and development of its structural planes.The structural plane of the rock is the product of a complicated tectonic process,which has the characteristics of a non-linear distribution.This paper uses the non-linear system science of chaos theory to research the non-linear spatial distribution patterns of the Qinling Tunnel area structural planes and construct a model of chaos.Advancement will be made beginning from the chaos characteristics of structural planes and the utilization of data to develop new surveying methods for the rock masses in the Qinling Tunnel area.(1) The structural series was constructed using the concept of time series analysis.In order to satisfy distance requirements such as time series,we formed an arrangement from surveys of the structural plane distances.From this,we could construct the structural plane dipdirect and rake series.From the combination of these three series we could determine the spatial orientation,dipdirect and rake of the structural plane.Using non-linear research methods including chaos theory,we studied the characteristics of the structural plan series and its structural formation.(2)Using surveying as the basis of our data,we utilized the non-linear science of chaos theory to research the characteristics of spatial distribution within the structural plane.The phenomenon of the existence of low frequency negative power within the power spectrum,the existence of the relationship between structural plane development and faults,and the phenomenon of the Kolmogoroc entropy structural plane distribution fractal environment greater than 0 all explain the existence of chaotic distribution within the structural plane.The Lyapunov index that tends towards 1,the principal component graded analysis,the appearance of the power spectrum and the disappearance of negative power also lead us to believe that the distribution of the structural plane is not completely chaotic,but is at the edge of chaos.(3) phase-space reconfiguration reveal the spatial distribution of non-linear law based on chaotic time series phase-space reconstruction theory.Using the CC method,the smallest methods of error prediction in the embedding dimension and delays the amount of space showing the distribution of the rock structure.Explicit embedding dimension of the is similar to the structure of the main rock face in the development of the average group number,and the main space delay structure is similar to the average interval between the chaos,such as the physical meaning of parameters.(4) Through the calculation of the amount of chaos in the structural plane,we establish the predicative chaos models for the structural plane locations and a chaotic occurrence predicted sequence structure prediction provides the basis for the decision to phase-space reconstruction of chaotic prediction method.The single-order local-region method algorithm is simple and easy to use,the sequence prediction accuracy not only depends on the embedded dimension of structure and spatial delay,it is also related to the closest related points,training sample size and estimations such as step size.(5) Based on the chaotic characteristics of the structural planes we predicted the surrounding rock mass classifications.Using the known classification of rock masses surrounding the structural planes as training samples,the structural plane samples as a prediction to the structure of the cutting weight density,the structure of surface spacing length of structural planes,and the orientation and inclination for the four predictor attribute factors according to the theory of excavation discretization data structure to basic Bayes’s classification,we predicted unknown samples of rock classification.The results of the engineering project proved to be feasible.This thesis uses the nonlinear systems of chaos theory as a scientific theoretical foundation,and uses basic knowledge from Bayesian classification as a means to predict the rock classification of the Xi’an-Ankang Railway Qinling Tunnel.The results not only provide technical support for the design and construction works,they also can be used to explore these issues a new way for other projects.更多还原