Traditional dynamic programming has the defects of consuming large computational resources and long computation time. This drawback is more obvious in complex models, such as multi-objective energy management strategies for turbo-electric hybrid propulsion systems. The existing modification of dynamic programming was mainly focused on the unit commitment problem instead of a large, complex engineering model. Therefore, this paper proposes State Machine-based Dynamic Programming (SM-DP), which aims to utilize expert knowledge to narrow down the selection space of policy actions, sacrificing very small optimality for faster computation. Simulation results demonstrate that the proposed method achieves a 14.84-fold computational speed advantage over conventional dynamic programming approaches, while maintaining solution quality with merely a 0.456% optimality reduction. In addition, SM-DP in this study achieves good results in both flight routes with harsh flight environments and environments with more refined maneuver decisions.