CLUSTER ANALYSIS FOR JOINT DATA OF THREE-DIMENSIONAL ROCK MASS USING DifFUZZY METHOD

Fuzzy clustering techniques are often used in the study of three-dimensional data sets,such as orientation data. Other methods are used in finding groups within directional data. But traditional methods usually meet difficulties when dealing with some complicated data sets. In this paper,a new DifFUZZY clustering method proposed by Ornella Cominetti and Anastasios Matzavinos for the complex data sets,is firstly applied to cluster the orientation data. The results show the follows. (1)DifFUZZY is a new simple and powerful method to realize the 3D-simulation of joint sets,which can automatically separate directional data into distinct clusters and get the dominated value of orientation data. (2)DifFUZZY does not require any advanced information about the orientation data,and it just randomly selects the initial values as the dominated orientation and then according to its own algorithm finally decide which points belong to a cluster and the finally dominated orientation values of the clusters. The suggested method makes it possible to analyse the local stability of rock slope quickly. (3)DifFUZZY can self-acting return the cluster number and on the principle of the membership value it can automatically decide the point lies in which cluster or not.