To effectively adapt to the dynamic characteristics of UAVs, network load variations, and diverse service requirements, we propose an innovative Q-learning based UAV ad hoc network hybrid media access control (MAC) protocol. This protocol combines time division multiple access (TDMA) protocol and statistical priority-based multiple access (SPMA) protocol, leveraging long short-term memory network (LSTM) for real-time channel load prediction and Q-learning for adaptive and intelligent MAC protocol switching. It automatically switches to TDMA protocol for light loads or low-latency transmission. For heavy loads or diverse services, it intelligently switches to SPMA protocol, reducing complexity and enhancing system flexibility based on channel load. Results demonstrate significant improvements in throughput and latency with the Q-learning based UAV ad hoc network hybrid MAC protocol compared to traditional SPMA protocols. It has the capability to intelligently select the appropriate MAC protocol in real-time, allowing it to adapt to varying loads and service requirements.