The analysis of discontinuities occurrence in rock mass is an important fundamental work for mechanical and hydraulic behaviors analysis of rock mass. Discontinuities with the same occurrence and different other properties have different mechanical and hydraulic behaviors. Therefore

there are defects in conventional method

due to that they partition discontinuities only according to their occurrence. A new method based on mutative scale chaos optimization algorithm is proposed for partitioning multivariate data for charactering discontinuities. The objective function is established based on the similarity measure between discontinuities; then

mutative scale chaos optimization algorithm is used to search the optimum solution. This new method can simultaneously divide discontinuities into several groups according to their multivariate properties. The accuracy of the new method is proved by artificial data. Finally

this method is applied to grouping of field discontinuity data with five properties; and reliable cluster results are obtained.