高超声速飞行器的调度离线预测控制

Scheduled offline model predictive control for a generic hypersonic vehicle

本文将调度预测控制的思想应用于离线鲁棒预测控制,设计了高超声速飞行器计算有效的调度离线预测控制器.首先在不同的平衡点离线设计一系列控制规则,实际实施时只需要在不同的控制器之间进行切换,避免进行在线优化,大幅度减少了在线计算时间.通过估计局部控制器的稳定域,保证了切换控制器的稳定性.另外在确保良好控制品质的同时,还能够保证所有输入和状态均在给定约束范围.仿真试验表明,提出的方法能实现速度和高度较大范围的指令跟踪,所有输入和状态均在给定约束范围内;相比于在线鲁棒预测控制方法,仿真运行时间减少,可以实现高超声速飞行器的实时控制.

A computationally efficient scheduled offiine model predictive controller is designed for a generic hypersonic vehicle with the offline robust model predictive algorithm based on the idea of the scheduled model predictive control. Local predictive controllers are firstly designed offline around different equilibrium points, and the real-time implementation includes only the simple switching among local controllers; which avoids the on-line optimization, and the on-line computation time is reduced greatly. Through estimating the stable region of the local controllers, the stability of the scheduled controller is guaranteed. Moreover, the predictive control allows the input and the state constraints to be considered explicitly in the design phase; hence, the control performance and the constraints of the input and the states are guaranteed simultaneously. Simulation results show that the altitude and the velocity can track the reference signals in a relatively large scope with the input and the states in the given ranges. Compared with the well-known online robust model predictive control, the simulation time is greatly reduced, and the desired real-time control for the hypersonic vehicle is attainable.