Abstract: Discrete fracture networks can be used to simulate a wide range of physical processes such as mechanical properties, seepage characteristics, and heat transfer in coal rock bodies, and have important applications in many fields. Discrete fracture network skeleton extraction essentially involves selecting the more important key fractures in the network and connecting them in series. In this paper, it proposes a method for extracting the skeleton of the fracture network based on the identification of key fractures, which firstly carries out the identification of key fractures, and on this basis, constructs the skeleton of the connected discrete fracture network. In terms of key fracture identification, this paper introduces the definitions of decomposition number and decomposition factor, improves the traditional K-shell algorithm, solves the problem of insufficient differentiation of K-shell algorithm on the importance of different fracture, and improves the identification effect of key fracture. On the basis of key fracture identification, this paper proposes an algorithm of skeleton extraction of discrete fracture networks, which ensures the connectivity between fractures while continuously selecting important fractures, and takes into account the in-plane and out-plane properties of the fracture network. Simulation experiments on the discrete slit network dataset show that the algorithm proposed in this paper can effectively achieve the skeleton extraction of discrete fracture networks.