Flight action recognition is crucial for segmenting load spectrum tasks in aviation engines. To address the shortcomings of traditional action recognition methods, a method is proposed based on the segmentation of helicopter flight into takeoff, cruising, and landing phases. The research introduces a complex action recognition approach refined from six fundamental actions: hovering, level flight, ascent, descent, turning, and rolling, while considering parameters such as altitude, rotation speed, and power. Through an in-depth analysis of the parameter variation characteristics of complex actions outlined in helicopter flight manuals, and incorporating insights from domain experts, the study emphasizes the parameter variation ranges of basic actions, the spatial combinations of complex actions, and their cumulative parameter changes. This has facilitated the development of a portable flight action recognition program. A simulation experiment was conducted using flight mission data from a certain type of helicopter. The results demonstrate that this method enables quick and highly accurate recognition of complex actions.