In the process of mine construction, deep-buried chambers often face the problem of high ground stress accumulation, which may lead to many engineering diseases, such as rockburst and large deformation of soft rock. In-situ stress, as the main motive force of deformation and destruction of underground rock mass and ore body, has great influence on rock mass deformation and engineering stability. In order to ensure the smooth progress of mine construction, it is very important to accurately grasp and analyze the in-situ stress parameters and their distribution characteristics in mining areas. In the exploration process of Wenquan mine field in Qinghe mining area, the in-situ stress in deep holes was tested by selecting appropriate drilling locations, and the in-situ stress parameters were obtained and their distribution characteristics were analyzed, which provided an important design basis for the next construction of the mine. Based on the measured results of in-situ stress caused by hydraulic fracturing in 10-4 borehole in Wenquan mine field, the in-situ stress parameters of surrounding rock in the test blank area of the mining area were estimated and obtained by using the improved Sheorey calculation model, and the BP neural network optimized by particle swarm optimization algorithm PSO was applied with high accuracy. On this basis, the rock burst phenomenon of hard surrounding rock and the deformation of soft surrounding rock that may occur during the excavation of + 1 500 m horizontal transportation roadway in mine were predicted. By analyzing the distribution characteristics of in-situ stress, possible engineering diseases can be effectively predicted, which can provide an important basis for the next engineering construction and roadway support design in the mine.