Due to the complex underground space environment in deep mines, transmission interruption and high transmission delay often be appeared in the process of communication transmission. To solve it, a large number of nodes need to be added which increases the communication transmission cost. We proposed a probabilistic routing algorithm based on message utility in coal mine. Under the existing conditions, we use the sociality of people to assist in decision-making transmission strategies. Firstly, based on the movement characteristics of underground nodes, the movement track similarity, node encounter duration and node similarity are defined; Secondly, the underground work nodes are classified into different communities according to their sociality, and different message forwarding strategies are used inside and outside the communities to improve transmission efficiency; Finally, by adding an ACK deletion mechanism to limit network redundant copies, network congestion is effectively avoided. The simulation results on One platform show that the algorithm can improve the success rate of message delivery and reducing the network delay in the mine scenario with poor communication environment.