Unmanned Aerial Vehicle (UAV) is attracting attention as a key element of the New Radio (NR) system as it can exceed the limits of the ground network to prepare for the rapidly increasing data usage and create new application services. However, high-altitude UAVs have a high Line-Of-Sight (LOS) probability, resulting in interference problems between adjacent cells in multi-cell networks. Therefore, this paper proposes an algorithm that selects the optimal beam to reduce the impact of interference and ensure high transmission efficiency. Specifically, the proposed algorithm consists of the process of building a Fingerprint Database (fingerprint database) and the process of selecting a load balancing beam. Simulations were conducted based on the downlink system for performance analysis, and signal-to-interference-plus-noise cumulative distribution function (SINR CDF) is used as performance analysis indicators.