<p>This article explores the utilization of multiple flying base stations (FBS) created by mounting cellular base station antennas on unmanned aerial vehicles (UAVs) to restore wireless services in scenarios where the terrestrial network has collapsed or is non-existent. The primary objective is to enhance cellular coverage while satisfying the quality-of-service (QoS) requirements of cellular users (CUs) during emergency scenarios with minimal FBS positioning. To achieve this goal, a Kmeans-based repair genetic algorithm (RGA) is proposed. Initially, considering limited resources and hardware availability for FBS, the maximum coverage radius of the FBS is estimated based on network parameters. Subsequently, the proposed approach is implemented in three steps. Firstly, K-means, a clustering technique based on machine learning (ML), is employed to initialize the stationing locations of FBSs. Secondly, a modified genetic algorithm that integrates a penalty term into the overall objective function for handling the constraint violation is proposed to obtain the optimal FBS stationing locations. Thirdly, a repair routine is designed to to mitigate constraint violations. Furthermore, to provide a comprehensive understanding, the algorithm’s performance is evaluated with two alternative techniques employed for the initialization of FBS stationing locations. Finally, the effectiveness of this proposed approach is demonstrated through simulation results.</p>