Unmanned aerial vehicle air combat decision-making method and system combining reinforcement learning and game theory

The invention discloses an unmanned aerial vehicle air combat decision-making method and system combining reinforcement learning and a game theory, and the method achieves the nonlinear modeling of a high-dimensional and continuous state and a strategy space of an unmanned aerial vehicle through deep reinforcement learning, shows high adaptability, and can respond to a complex and dynamic environment in real time. In the training construction process of the decision network, the air combat situation is divided according to the sight angles of the unmanned aerial vehicles of the two air combat parties, and meanwhile, the unmanned aerial vehicle dual-network confrontation training is designed, so that the enemy unmanned aerial vehicles make decisions by using the previously trained model, and the decision performance of the unmanned aerial vehicles is further improved. According to the method, the air combat process of the unmanned aerial vehicles of the two parties is regarded as a two-person zero sum Markov g