Abstract: The rock mass discontinuity significantly affects the mechanical and hydraulic properties of rock. When analyzing and evaluating the discontinuity characteristics of rock mass,it is necessary to divide discontinuities into the same group based on the similarity of some properties. This helps to understand the development of the discontinuities of different properties in the rock mass. In the field of engineering,the Fuzzy C-Means(FCM)clustering algorithm is a commonly employed method for this purpose. However,the FCM algorithm has inherent shortcomings such as being sensitive to initial center selections and susceptibility to local optima. In this paper,a mixture model method for dividing the dominant partitioning of rock mass discontinuity orientation,based on a genetic simulated annealing algorithm and Fuzzy C-Means clustering algorithm(GSA-FCM),is proposed. This method offers a straightforward principle and swift computational performance,and the Metropolis criterion of the simulated annealing is integrated into the genetic algorithm,utilizing the Genetic Simulated Annealing Algorithm to determine the clustering centers for discontinuities,subsequently optimizing the results produced by the FCM algorithm. This method aims to mitigate the shortcomings associated with traditional FCM clustering,primarily addressing the impact of initial center selections and the risk of converging to suboptimal solutions. Analysis of discontinuity orientation data,generated through computer simulations,demonstrates that the proposed GSA-FCM mixture model method exhibits clear advantages over conventional FCM clustering approaches. Finally,applying the GSA-FCM mixture model method to real measured discontinuity orientation data from the Nujiang Maji Hydropower Station in Yunnan,China,the results show that this method achieves high clustering accuracy and accurate grouping results,and is highly suitable for engineering applications.