引用本文:孙启超,徐娟娟,张焕水.高超声速飞行器再入轨迹跟踪的块压缩模型预测控制方法[J].控制理论与应用,2026,43(4):814~820.[点击复制]
SUN Qi-chao,XU Juan-juan,ZHANG Huan-shui.Block condensing model predictive control for reentry trajectory tracking of hypersonic vehicle[J].Control Theory & Applications,2026,43(4):814~820.[点击复制]
高超声速飞行器再入轨迹跟踪的块压缩模型预测控制方法
Block condensing model predictive control for reentry trajectory tracking of hypersonic vehicle
摘要点击 153  全文点击 21  投稿时间：2024-01-31  修订日期：2024-12-20
查看全文  查看/发表评论  下载PDF阅读器   HTML
DOI编号  10.7641/CTA.2024.40089
  2026,43(4):814-820
中文关键词  高超声速飞行器  轨迹跟踪  模型预测控制  块压缩  序列二次规划
英文关键词  hypersonic vehicles  trajectory tracking  model predictive control  block-condensing  sequential quadratic programming
基金项目  国家自然科学基金创新研究群体项目(61821004)资助.
作者单位E-mail
孙启超 山东大学 控制科学与工程学院 1047663353@qq.com 
徐娟娟 山东大学 控制科学与工程学院  
张焕水* 山东大学 控制科学与工程学院 hszhang@sdu.edu.cn 
中文摘要
      本文针对高超声速飞行器的再入轨迹跟踪控制问题, 考虑了非线性模型预测控制的计算加速方法. 首先采 用序列二次规划算法, 将非线性最优控制问题转化为迭代差值的线性最优控制问题, 再应用块压缩方法, 将长时域 问题转化为多段较短区间内的最优控制问题. 利用对偶特性推导了由拉格朗日乘子优化问题和分段最优控制问题 组成的双层迭代结构. 由于拉格朗日乘子迭代优化, 相邻的最优问题解耦, 可以并行计算. 最后, 对控制和拉格朗日 乘子同时进行热启动, 减少迭代次数, 提高求解速度. 仿真结果显示, 在保证控制效果的前提下, 控制器求解速度显 著提高.
英文摘要
      In this paper, a computational acceleration method for nonlinear model predictive control is considered to track trajectories for the reentry of hypersonic vehicles. First, a sequential quadratic programming algorithm is adopted to transform the nonlinear optimal control problem into a linear optimal control problem with iterative differences. Then, a block condensing method is applied to transform the long horizon problem into an optimal control problem with multiple short intervals. The principle of duality is exploited to derive a two-layer iterative structure that consist of a Lagrange multiplier optimization problem and segmented optimal control problems. As the nearby optimal problems is decoupled by the iterative optimized Lagrange multiplier, they are computed in parallel. Finally, the control and Lagrange multipliers are warm-started simultaneously to reduce the number of iterations and improve the solution speed. The simulation results show that the solving speed of the controller is significantly enhanced while maintaining the effective control performance.