This study explores the deployment of aerial base stations carried by UAVs to assist terrestrial cellular networks to enhance communication coverage. This approach offers flexibility, speed, and cost-effectiveness. Dynamic service requirements are one of the critical factors affecting the deployment of aerial base stations. The current research focus is to fully consider the constantly changing service requirements and achieve more intelligent and efficient deployment of aerial base stations. The paper investigates the impact of user mobility on networks organized by aerial base stations. The user mobility causes dynamic changes in wireless network services, which may lead to performance loss. Achieving optimal performance requires an algorithm that can quickly adjust the location of aerial base stations based on real-time distribution changes. Therefore, based on the Deep Q-network (DQN) deep reinforcement learning framework, we propose an algorithm for deploying aerial base stations in dynamic service scenarios. The simulation results demonstrate that the proposed method provides an effective deployment strategy for improving wireless network coverage in dynamic service scenarios.