针对旋翼无人机编队飞行过程中高速穿越狭窄通道并尽可能保持相近编队构型的问题，提出一种多阶段无人机编 队队形安全变换策略。首先将无人机编队中心看成虚拟领导者，根据无人机的运动学约束，通过混合整数线性规划规划出 一条平滑、高速、距离障碍物最远的虚拟领导者轨迹以及编队整体的安全飞行空间。然后根据编队安全飞行空间和虚拟领 导者轨迹求解一个无外部和内部碰撞，且相较初始构型改变较少的编队构型。之后基于生成的编队构型和虚拟领导者轨 迹，分布式计算每个无人机的期望轨迹，并提出一种结合非线性优化与协同时间重分配的分布式算法，对各无人机轨迹进 行优化，使其在满足运动学约束的前提下，实现高速飞行与编队保持。仿真实验结果表明：所提算法能使无人机编队在满 足运动学约束的情况下，高速、安全穿越狭窄通道，同时尽可能保持相近的编队构型。

To address the problem of enabling a multirotor UAV formation to traverse narrow passages at high speed while maintaining a formation configuration as close as possible to the initial one, we propose a multi-stage safe formation transformation strategy. First, the center of the UAV formation is treated as a virtual leader. Considering the kinematic constraints of the UAVs, a smooth, high-speed, and obstacle-avoiding trajectory for the virtual leader, along with a safe flight corridor for the entire formation, is generated using mixed-integer linear programming. Then, based on the safe flight corridor and the virtual leader's trajectory, a new formation configuration is computed that avoids both external and internal collisions while preserving the initial configuration as much as possible. Subsequently, with the obtained formation configuration and the virtual leader trajectory, the desired trajectory of each UAV is computed in a distributed manner. A distributed algorithm that combines nonlinear optimization with cooperative time reallocation is further proposed to optimize the UAV trajectories such that high-speed flight and formation maintenance are achieved under kinematic constraints. Simulation results demonstrate that the proposed method enables the UAV formation to traverse narrow passages safely and at high speed while maintaining a configuration close to the initial one.