基于统计参数的二维节理粗糙度系数非线性确定方法

发布时间：2018-11-17 11:47

【摘要】：岩体节理粗糙度系数JRC与其统计参数之间具有复杂的非线性关系,单一统计参数因存在对结构面形貌描述的片面性,从而导致JRC计算结果的可靠性较低。从结构面起伏角、起伏高度及其分布特征的角度选取了平均起伏角avei、平均相对起伏度H/L、起伏角标准偏差iSD和起伏高度标准偏差hSD 4个参数共同反映结构面形貌。以已知JRC试验反算值的102条结构面剖面线作为样本数据对支持向量机进行训练,构建JRC与所选取的统计参数之间的非线性映射关系,建立了JRC支持向量回归(SVR)预测模型,并通过Barton标准剖面线的JRC预测值与试验反算值的对比证明了模型的可靠性。以三峡库区秭归县马家沟滑坡所处地层岩体结构面为例,基于三维激光扫描试验获取了其表面形貌数据并建立了三维形貌模型,开展室内直剪试验反算得到了其JRC。实例JRC的计算结果表明,SVR模型预测结果与试验反算值的相对误差仅为4.5%,不同统计参数回归关系式对于相同剖面线的估算结果存在较大差异,表明基于所选取的统计参数,采用SVR模型预测得到的JRC更加可靠。该方法为JRC的定量确定提供了新思路。
[Abstract]:There is a complex nonlinear relationship between rock joint roughness coefficient JRC and its statistical parameters. Because of the one-sidedness of the single statistical parameter describing the morphology of the structural plane, the reliability of the JRC calculation results is low. From the angle of fluctuation angle, height and distribution of structure surface, four parameters, mean relative fluctuation degree H / L, standard deviation of fluctuation angle iSD and standard deviation of fluctuation height hSD, are selected to reflect the morphology of structure surface. The support vector machine (SVM) is trained with 102 structural plane profiles of known inverse value of JRC test as sample data. The nonlinear mapping relationship between JRC and selected statistical parameters is constructed, and the JRC support vector regression (SVR) prediction model is established. The reliability of the model is proved by comparing the JRC prediction value of Barton standard profile with the experimental backcalculation value. Taking the stratum rock mass structure plane of Majiagou landslide in Zigui County of the three Gorges Reservoir area as an example, the surface topography data were obtained based on the 3D laser scanning test and the three-dimensional topography model was established. The JRC. was obtained by using the indoor direct shear test. The calculation results of JRC show that the relative error between the prediction result of SVR model and the back calculation value of the experiment is only 4.5, and there are great differences between the different regression equations of statistical parameters and the estimation results of the same profile line. It shows that the JRC predicted by SVR model is more reliable based on the selected statistical parameters. This method provides a new idea for quantitative determination of JRC.
【作者单位】：中国地质大学(武汉)工程学院;
【基金】：国家自然科学基金资助项目(No.41372310) 中国博士后基金项目(No.2015M570671) 中国地质大学(武汉)中央高校基本科研业务费专项资金资助项目(No.1610491T07)~~
【分类号】：TU45

【相似文献】