关键词: 轨迹优化, 数据驱动, 稀疏多项式混沌展开, 助推滑翔飞行器, 序列凸优化

Abstract: Most trajectory planning for high-speed aircraft is based on deterministic trajectory optimization models, which struggle to handle the multi-source uncertainties present in actual flight scenarios, leading to insufficient robustness in the planned trajectories. To address this issue, this paper first constructs a robust trajectory optimization model. Then, a data-driven method is used to build a sparse polynomial chaos expansion surrogate model, thereby transforming the double loop nested uncertain trajectory optimization problem into a dimension-expanded deterministic trajectory optimization problem. Finally, a sequential convex optimization-based deterministic trajectory optimization solution strategy is employed to solve this dimension-expanded deterministic optimization problem and obtain the optimized trajectory. Since the proposed method is a data-driven uncertain trajectory optimization design approach, it does not require prior knowledge of the full probability distribution of uncertain variables, avoiding the impact of incorrect uncertainty assumptions on the optimization results. Trajectory optimization results of a specific boost-glide high-speed aircraft demonstrate that the proposed method can mitigate the influence of multi-source uncertainties on the optimization results. Moreover, it has advantages of considerable accuracy and computational efficency when compared with existing uncertain trajectory optimization methods for high-speed aircraft.

Key words: trajectory optimization, data-driven, sparse polynomial chaos expansion, boost-glide high-speed aircraft, sequential convex optimization