With the gradual unmanned, intelligent, and clustered development of aircraft, the air battlefield is gradually entering the era of intelligent air combat. Major aviation powers such as the United States and China, as well as related research institutions, are also focusing on exploration and research of key technologies for intelligent air combat. Deep reinforcement learning combines the perceptual ability of deep learning with the decision-making ability of reinforcement learning, demonstrating significant advantages in the emergence of air combat capabilities. This article, based on the urgent needs of intelligent air combat development, analyzes and summarizes the mainstream algorithms in the field of deep reinforcement learning, and explores the points of integration with the air combat field. From the perspective of algorithm implementation, it identifies key technologies of deep reinforcement learning in air combat. By sorting out the current cutting-edge technological achievements in the field of air combat, it is concluded that the future research on deep reinforcement learning will develop from single-to-single air combat to cluster air combat. Finally, the challenges algorithm faces are proposed, providing reference and guidance for the development of intelligent algorithms in intelligent air combat.