多参数统计均质区划分——以马吉水电站为例

Multivariate method for identifying structural domain boundaries:an example from Maji Hydropower Station in China

引入秩和检验非参数统计方法,建立不连续面岩体统计均质区评价模型。文中考虑了不连续面产状、迹长两个参数,在随机数学的基础上提出了一种不连续面岩体统计均质区划分的新方法,即利用秩和检验计算研究区样本之间的相关性,从而合理划分不连续面岩体统计均质区。选取云南省马吉水电站左岸坝肩某两高程4个平硐作为研究对象,首先将产状数据用其法向极点与单位圆上半球被划分为100个等面积四边形小方格的交点来表示;其次,根据迹长从小到大对落入小方格的极点进行编秩;最后,基于秩和检验来分析不连续面落入相应小方格的频率。研究结果表明:PD201和PD251以及PD251和PD253不具有统计相似性,而PD241和PD253、PD201和PD241在不同的比较区域下得到不同的均质区评价结果,这说明统计均质区存在尺寸效应。较Miller仅考虑不连续面产状,本文同时考虑了产状和迹长两个因子,结果更加合理。此外,该方法可以应用到小样本不连续面统计均质区划分中。

The Wilcoxon rank sum nonparametric statistical test is introduced to establish an evaluation model for identifying structural domain boundaries in jointed rock masses. A new method for identifying structural domain boundaries considering discontinuity orientation and trace length is proposed on the basis of random mathematics. The Wilcoxon rank sum test is used to calculate the correlation between samples collected from four adjacent adits at different elevations of Maji Hydropower Station at the left bank of Nujiang River in Yunnan Province of China so as to rationally divide statistical homogeneous areas of discontinuous rock masses. First, the Schmidt plots are divided into 100 windows with approximately equal areas. Second, the poles occurring in each corresponding window are arranged in ascending order with respect to their trace lengths. Finally, the Wilcoxon rank sum test is used to identify the homogeneity of structural populations by analyzing the discontinuity sequence. The results indicate that, PD201 and PD251, PD251 and PD253 are situated in different structural domains, while PD241 and PD253, PD201 and PD241 in different areas have different homogeneous area evaluation results, indicating that the statistical homogeneous area has size effect. The Miller＇s method only considers discontinuity orientation, while the new method is more reasonable becomes it considers both the orientation and the trace length. The other advantage of the new method is that it can be employed to evaluate statistically homogeneous regions with small sample sizes.