Course learning-based unmanned aerial vehicle air game confrontation solution method

The invention discloses a course learning-based unmanned aerial vehicle air game confrontation solution method. The method comprises the following steps of: (1) constructing an analogue simulation environment; (2) collecting real trajectory data of a pilot for controlling an aircraft, and performing course target classification on the trajectory data according to maneuvering motion difficulty; (3) for a trajectory under a specified course target, optimizing the similarity between a trajectory generated by a strategy model and an expert trajectory through imitation learning; (4) obtaining a pre-trained unmanned aerial vehicle strategy model; (5) based on the pre-trained unmanned aerial vehicle strategy model, creating unmanned aerial vehicle intelligent agents of a friend and a foe in a simulator; (6) making the unmanned aerial vehicles obtain observation at the current moment in the simulator; (7) making the unmanned aerial vehicles interact with a simulated environment, modeling the task of confrontation bet