There are shortcomings in the deployment of power emergency communication base stations that only rely on human subjective experience. Therefore, considering the influence of buildings on the shielding of base station signals, the research problems are graphically described based on the return information of individual soldiers to the disaster area. In addition, a two-layer optimization model for the location of power emergency communication base stations is proposed. The upper layer model aims to minimize the signal transmission loss between base stations. Within the delineated optional location range of base stations, the particle swarm algorithm is used to optimize the pre-selection of base stations for the nonlinear model. The lower layer model aims to maximize the distance between the signal transmission path and the building. A support vector machine is used to optimize the final position of the base station, and this is transmitted back to the upper layer to calculate the signal transmission loss. The optimal deployment position of the base station is obtained through iterative optimization. The feasibility and effectiveness of the proposed method are verified by comparing with the single-layer model in an example. The work can provide a technical reference for the construction of a power emergency communication network after major natural disasters. This work is supported by the National Science and Technology Major Project of China (No. 2020YFC0827001).