The problem of knowledge acquisition for the air combat of fighter is studied, and a way to acquire knowledge from massive flight parameters data is put forward. First, expert system of air combat is built; then for flight maneuver knowledge extraction, a method based on grey wolf optimization (GWO) is proposed, in order to obtain simple and effective knowledge, and design a new evaluation function of the GWO algorithm. Finally, the simulation results of horizontal right turn and single plane loop verify that the knowledge extraction method of flight maneuver is effective and feasible.