对某矿区岩体节理的几何参数进行了现场勘测,获得节理产状、间距等原始数据。根据赤平投影原理,对原始数据进行数据转换、成图显示,得到节理分组的定性信息。运用聚类分析的方法对节理进行了快速聚类分组。根据离散系数来评价节理分组的可靠性,并选择最佳分组方案,得到节理分组的初步定量结果。对离散节理数据进行了优化,剔除了离散的随机节理,获得节理分组的定量信息。结果表明超多簇中心离散优化技术是快速聚类分析的有效补充,可以有效剔除节理分组信息中的随机离散信息。 The geometrical parameters of rock mass joints in a mining area were surveyed on site, and the original data such as joint shape and spacing were obtained. According to the principle of stereoprojection, the data of the original data are transformed and plotted to obtain the qualitative information of the joint grouping. Using the method of cluster analysis, the joints are clustered quickly. According to the discrete coefficient, the reliability of joint grouping is evaluated and the optimal grouping scheme is selected to obtain the preliminary quantitative results of joint grouping. The discrete joint data is optimized, the discrete random joint is eliminated, and the quantitative information of the joint grouping is obtained. The results show that the super cluster center discrete optimization technique is an effective complement to rapid cluster analysis, which can effectively eliminate random discrete information in the joint grouping information.