In complex environments, the location of the unmanned aerial vehicle (UAV) swarm airborne base station is not precisely known. This paper focuses on the optimal base station selection of UAV airborne base station positioning system with base station position error. For the propagation of base station position error, this paper uses an estimation method based on belief propagation to improve localization accuracy. This paper proposes a UAV base airborne station selection criterion based on the Fisher information matrix (FIM). The base station selection problem is described as a nonconvex optimization problem. The problem is relaxed as the convex semidefinite program by the Semi-Definite Relaxation (SDR). Simulation results show that the proposed evaluation criterion effectively reduces the running time of base station selection. Compared with other base station selection methods, the proposed method has better location accuracy in the scene with base station position error.