中文摘要：

岩石结构面的代表性取样是开展不同尺寸模型结构面直剪试验的前提条件，直接影响到结构面抗剪强度尺寸效应试验结果的有效性和可靠性。通过对取样尺寸为10 cm的18组岩石结构面粗糙度系数（JRC）的测量和统计分析，得到该尺寸岩石结构面试样的粗糙度系数分布规律，表明模型结构面取样代表性评价的必要性。基于结构面粗糙度系数非均一性分布特点，以及对简单随机取样法和分层取样法2种概率取样方法的比较分析，提出基于分层取样法的岩石模型结构面代表性取样方法，该方法采用JRC四分位法确定各层样本的数量比例，再利用分层取样公式计算样本的取样数量。通过岩体结构面抗剪强度的力学机制分析，对试样取样代表性和力学可靠性进行校验，结果表明，采用该取样方法能较好地代表该尺寸的表面形态分布，取样结果也较好地满足力学试验的统计精度要求。

英文摘要：

Representative sampling of rock joints is the precondition for direct shear test of the structural plane models with different sizes,which directly affects the effectiveness and reliability of the experimental result of shear strength scale effect. The joint roughness coefficient（JRC） distribution regularity is obtained by analyzing the statistical results of JRC distribution of 18 groups of rock joints with size of 10 cm. The statistical results indicate the necessity of representative evaluation for structural plane models. According to the distribution features of the nonuniformity of JRC,and the comparative analysis between simple random sampling method and stratified sampling method,the representative sampling method of structural plane model based on stratified sampling is proposed. The stratified proportions are determined by the quartile method of JRC. The sampling quantity is calculated through the stratified sampling formula. The sampling representativeness and mechanical reliability are inspected by the mechanical mechanism analysis of rock joints shear strength. The results show that the sampling method can represent the distribution of surface morphology for the certain size;and the sampling result also preferably meets the accuracy requirement of mechanical experiment.