To enable unmanned aerial vehicles (UAVs) to assess battlefield situations and generate corresponding maneuver commands, an intelligent decision-making algorithm for 1v1 UAV air combat is proposed. Firstly, a UAV flight trajectory prediction algorithm is designed to calculate the flight trajectory of the aircraft over a period of time. Secondly, a UAV state value function is designed, which can calculate the state value of the aircraft after this prediction to achieve the role of rapid rollout. The rapid rollout combat simulation is used to create a dataset, establish a Monte Carlo search tree, and calculate the probability of aircraft actions. The eight action combinations with higher probabilities are selected as candidate action sets for reinforcement learning. Considering the computational time constraints of the algorithm in UAV air combat scenarios, rapid rollout is used to prune the Monte Carlo search tree. Then, the reinforcement learning unit is utilized to generate the current aircraft action strategy. Finally, rules are established for single-agent and dual-agent combat simulations. The simulation results show that o Our UAV can autonomously make decisions with a higher probability of success based on the states of both our and enemy UAVs.