Aircraft maneuver recognition is an important basis for flight action evaluation, flight load spectrum research, flight simulation and other research work. Aiming at overcoming the disadvantage that most of the existing maneuvers are based on the existing standard actions, combined with the important points of the time sequence, an adaptive maneuver recognition method based on horizontal and vertical plane flight trajectory projection merging is proposed. This method extracts the maneuvers in the flight data according to the trend of the two-dimensional plane trajectory data and combines and refines the maneuvers with the idea of maneuvers splicing. A maneuver recognition case study is carried out with complete aircraft takeoff and landing data. It is found that this proposed method can enhance the recognition efficiency of flight maneuvers and the division efficiency.