Abstract: Joints are widely present in rock masses,and their development affects the stability and seepage characteristics of the rock masses. Due to the large number of joints,grouping is currently required for their study. Traditional grouping methods, such as relying on rose diagrams and density maps such as poles,cannot determine the specific data of each group of joints and have limited effect on grouping of discrete points. The contemporary clustering algorithm using machine learning also suffers from the deficiency that the number of clusters selected affects the grouping effect. In view of this,a joints' orientation clustering program (JOCP) based on an improved fuzzy clustering algorithm was developed on the MATLAB platform. JOCP takes the original coordinate data and the target number of clusters as the input,and the nodal yield data,the cluster centers,the distribution of clustering results and the validity index as the output. The program is used in the analysis of a thousand slope nodal data in Dalian,and the results prove that the program can improve the grouping certainty and achieve the objective and accurate grouping effect. This program can provide technical support for geological exploration,disaster prediction and other fields.

Key words: rock mass,nodal grouping,cluster analysis,program development,improved fuzzy algorithm