基于岩石微组构采用多元回归和人工智能技术估算岩石抗压强度

发布时间：2018-05-21 04:41

本文选题：微观特征 + 抗压强度　；参考：《西南交通大学》2014年博士论文

【摘要】：单轴抗压强度(UCS)是完整岩石的最重要参数之一,它通常用于地面和地下各种工程设计中。但是,有时由于岩石各向异性以及岩石结构和构造的复杂性,很难测得理想的数据。因此,采用经验方法来估计岩石的单轴抗压强度参数具有一定的实用性和经济性。基于此,本研究的目的是吸引工程地质学者采用较稳定的岩石微观岩相特征预测完整岩石的单轴抗压强度,研究成果在当前相关领域具有创新性。当岩石存在较明显的变形时,岩石的微观特征对于岩石的破坏形式的非常重要的影响,尽管,有时候在对岩石特性的初步分析时经常忽略这些内容。本课题采用人工神经网络(ANNS)和模糊系统(FIS)以及多变量回归(MR)方法研究了角岩的各向异性以及微观特征与单轴抗压强度(UCS)的关系。斜长角岩取自四川省理县老君沟区域的岩石边坡,所取岩石具有典型的结构构造,其力学特性分析对本区域岩石工程设计具有重要意义。在对岩相特征的定量分析和岩石强度参数测试的基础上,采用SPSS V.19.0对数据进行了全面的统计分析,结果表明矿物晶粒大小、形状、石英含量是影响单轴抗压强度(UCS)的主要因素。相对于早先得到的岩石构造对岩石单轴抗压强度(UCS)影响的模型,本研究选取了相对较少的因素,但是这些已经足够构建估算此类岩石型的单轴抗压强度(UCS)。为了评估模型的实用性,选取相关系数(R),方差(VAF)均方根(RMSE)三个指标对模型进行对比分析,结果表明虽然三个模型都可以可靠的预测岩石的单轴抗压强度(UCS),但ANN模型表现最优。因此,在缺乏相关地质资料的基础上,采用岩石的微观特征可以得到较满意的岩石的强度和变形参数。但,ANNs模型与其它两个模型相比,其最主要的问题是在输入输出数据之间关系的不透明性。此外,本次研究了岩石面理方位对岩石强度和变形的影响,这些因素对当地工程设计具有重要意义。为了测试面理方位对岩石强度和变形的影响,根据面理角度(β=0°,30°,60°和90°)进行取样,实验数据表明不同面理方位对岩石参数影响显著。平行面理的纵波和剪切波波速明显大于其他方向的测试值。岩石强度的各向异性比值系数在0.96-1.47之间波动,并不呈现明显的线性特征。而且,而弹性变形试验表明,弹性变形性能与角岩的微观结构、各向异性特征没有关系。不过,岩石的杨氏模量与岩石层理的方向β的关系具有“U型”和“递减型”关系。因此,建议进一步研究此类岩性的各向异性特征。
[Abstract]:Uniaxial compressive strength (UCS) is one of the most important parameters of intact rock, which is usually used in surface and underground engineering design. However, it is sometimes difficult to obtain ideal data due to rock anisotropy and the complexity of rock structure and structure. Therefore, it is practical and economical to estimate uniaxial compressive strength parameters by empirical method. Based on this, the purpose of this study is to attract engineering geologists to predict the uniaxial compressive strength of intact rock by using relatively stable microfacies characteristics of rock. The research results are innovative in relevant fields at present. When the rock has obvious deformation, the microcosmic characteristics of the rock have a very important influence on the failure form of the rock, although sometimes these contents are often ignored in the preliminary analysis of the rock characteristics. In this paper, artificial neural network (Ann), fuzzy system (fish) and multivariate regression (MRM) methods are used to study the relationship between the anisotropy and microscopic characteristics of kernels and uniaxial compressive strength (UCSs). The diagonal rock is taken from the rock slope of Laojungou area, Lixian County, Sichuan Province. The rock obtained is of typical structural structure, and the analysis of its mechanical properties is of great significance to the rock engineering design in this region. On the basis of quantitative analysis of lithofacies characteristics and measurement of rock strength parameters, SPSS V.19.0 is used to make a comprehensive statistical analysis of the data. The results show that the size, shape and quartz content of mineral grains are the main factors affecting uniaxial compressive strength. Compared with the previous models of the influence of rock structure on uniaxial compressive strength (UCSs) of rock, relatively few factors are selected in this study, but these are enough to construct UCSs to estimate the uniaxial compressive strength of these rock types. In order to evaluate the practicability of the model, the correlation coefficient (R) and the mean square root (RMSE) are selected to compare and analyze the model. The results show that although all the three models can reliably predict the uniaxial compressive strength of rock, the ANN model is the best. Therefore, based on the lack of relevant geological data, satisfactory rock strength and deformation parameters can be obtained by using the microscopic characteristics of rocks. However, compared with the other two models, the main problem of ANNs model is the opacity of the relationship between input and output data. In addition, the influence of rock face azimuth on rock strength and deformation is studied, which is of great significance to local engineering design. In order to test the effect of face orientation on rock strength and deformation, samples were taken according to the plane angle (尾 0 掳~ 30 掳~ 60 掳and 90 掳). The experimental data show that different plane azimuth has a significant effect on rock parameters. The velocities of longitudinal and shear waves in parallel plane are obviously higher than those in other directions. The anisotropic ratio coefficient of rock strength fluctuates between 0.96-1.47 and does not show obvious linear characteristics. Moreover, the elastic deformation test shows that the elastic deformation performance is not related to the microstructure and anisotropy of the kernels. However, the relationship between the Young's modulus of rock and the direction 尾 of lithology has the relation of "U type" and "decreasing type". Therefore, it is suggested to further study the anisotropic characteristics of such lithology.
【学位授予单位】：西南交通大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TU45

【相似文献】