外弹道加权最小一乘模型研究与应用  

Study of Weighted Least-absolute Algorithm and Applications in Calculating Trajectories of Spacecraft

作　　者：徐小辉[1] 郭小红[1] 赵树强[1] 

机构地区：[1]西安卫星测控中心,陕西西安710043

出　　处：《飞行器测控学报》2008年第2期85-88,共4页Journal of Spacecraft TT&C Technology

摘　　要：工程技术领域中常常要求算法具有较强的稳健性。本文讨论了最小一乘和加权最小二乘相比于传统最小二乘的优势，根据实际工程需要，引入自变量权系数设置方式，并采用遗传算法提高算法的收敛速度，提出适用于工程计算的加权最小一乘模型。通过弹道数据的仿真计算，加权最小一乘算法能够克服异常数据所带来的影响，减小各种相对误差水平，表现出较强的稳健性，验证了该算法的有效性和正确性。High-level algorithm robustness is often required in engineering. This paper analyses the advantage of the least-absolute algorithm and weighted least-square algorithm to normal least-square algorithm. A weighted least-absolute algorithm model is proposed. The new algorithm adopts weigh coefficient setting method according to independent variables and genetic algorithm to increase convergence speed. It eliminates the influence resulting from outliers and reduces sensitivity to stochastic errors. Robustness of the algorithm and its validity is proven in application.

关 键 词：稳健性 加权最小一乘 遗传算法 异常值 

分 类 号：V557[航空宇航科学与技术—人机与环境工程]