Abstract: The narrow and branched structure of coal mine roadways makes it difficult for 5G signals to achieve full coverage, and the high path loss within the roadways limits wireless signal transmission. Therefore, site selection planning for coal mine roadway base stations is necessary. Existing methods typically focus on maximizing the coverage of visual range, neglecting the site selection problem in non-visual range. To address this issue, a coal mine roadway base station site selection method based on the ray-tracing path loss model is proposed. The coverage radius of base stations in rectangular cross-section roadways was determined using the ray-tracing path loss model. A digital elevation model of the roadway was constructed, and based on this, the objective function for base station location optimization was defined. The optimal coverage rate was achieved by minimizing the number of base stations and optimizing their locations. A genetic algorithm was employed to optimize the base station locations, thereby determining the optimal positions. Simulation results showed that with a site selection plan using 14 base stations, the network coverage rate was 91.2%, with an error of only 2.4% compared to numerical calculation results. Field measurement results in the coal mine roadway showed that the received signal power was slightly lower than the simulated results, but a signal strength of −80 dBm provided a coverage radius of 200 m, validating the effectiveness of the ray-tracing path loss model.