摘要
为提升航天发射任务弹道生成的响应效率与适应能力,分析发射任务关键因素建立弹道库,提出基于相似性指标的弹道匹配与快速生成方法。针对传统网格划分法因维度灾难导致采样效率低下,以及牛顿迭代等传统优化方法初值依赖性强、易陷局部最优的问题,采用拉丁超立方采样生成多维参数组合,构建覆盖多维环境变量的弹道库,降低高维弹道库构建的计算复杂度;结合改进差分进化算法与序列二次规划方法,构建混合优化框架进行弹道规划,融合前者全局搜索能力与后者局部快速收敛性,有效解决传统梯度类方法在复杂约束下收敛不稳定、解质量差的问题。针对航天发射任务,设计相似性匹配准则,实现相似任务的快速检索与弹道调用。数值仿真表明,该方法在优化质量与响应效率上均优于传统策略,能显著缩短弹道生成时间,为快速响应航天发射任务提供可靠支撑。
To enhance the responsiveness and adaptability of trajectory generation for aerospace launch missions,key mission factors are analyzed to establish a trajectory library,and a trajectory matching and rapid generation method based on similarity metrics is proposed.Addressing the inefficiency caused by the curse of dimensionality in traditional grid methods and the strong initial-value dependence,local optima convergence issues of conventional optimization approaches,Latin Hypercube Sampling(LHS)is employed to generate multi-dimensional parameter combinations.A trajectory library covering multi-dimensional environmental variables is constructed,reducing computational complexity in high-dimensional library development.A hybrid optimization framework integrating an improved Differential Evolution(DE)algorithm and Sequential Quadratic Programming(SQP)is developed for trajectory planning.By fusing DE’s global search capability with SQP’s local rapid convergence,unstable convergence and poor solution quality of traditional gradient-based methods under complex constraints are effectively resolved.For aerospace launch missions,a similarity matching criterion is designed to enable rapid retrieval of analogous missions and trajectory invocation.Numerical simulations demonstrate that the proposed method outperforms conventional strategies in optimization quality and response efficiency,with trajectory generation time significantly shortened,providing reliable support for rapid-response launch operations.
作者
郭晶晶
王建华
徐鹏
于沫尧
傅嘉威
GUO Jingjing;WANG Jianhua;YU Moyao;XU Peng;FU Jiawei(Department of Aerospace Science and Technology,Space Engineering University,Beijing 101416,China)
出处
《弹箭与制导学报》
北大核心
2025年第5期943-952,共10页
Journal of Projectiles,Rockets,Missiles and Guidance
基金
军委科技委基金(TKJSLLYJKT2024-1-A-028)
军委科技委基金(JH2024B010200-002-034)
某部队基金(BLB24L012)。
关键词
弹道库
拉丁超立方采样
差分进化算法
序列二次规划
trajectory database
latin hypercube sampling
differential Evolution algorithm
sequential quadratic programming
