As mobile communication networks expand rapidly, optimizing their performance becomes crucial. This paper investigates the use of reinforcement learning (RL) to dynamically manage base stations (BS) and sector coverage in response to real-time network demands. Our RL models learn to power down and reactivate BS based on user demand and adjust sector coverage to maintain service quality. Validated using real-world data, our approach demonstrates improvements in adaptability and performance. The results highlight the effectiveness of RL in balancing operational efficiency with service quality and its resilience to evolving network conditions.