基于B样条的稳健回归算法  

Robust Regression Algorithm Based on B-spline

作　　者：梁帆[1] 潘高田[1] 王庆权 解国栋[3] 黄一斌[1] 

机构地区：[1]装甲兵工程学院装备指挥与管理系,北京100072 [2]863部队,吉林白城137001 [3]装甲兵工程学院控制工程系,北京100072

出　　处：《装甲兵工程学院学报》2013年第1期97-100,共4页Journal of Academy of Armored Force Engineering

基　　金：军队科研计划项目

摘　　要：利用B样条基构建了拟合函数模型。针对观测数据中存在野值的情形,选取了稳健函数进行稳健回归,并通过迭代再加权法实现了拟合函数模型的求解。试验结果表明:在数据中含有野值的情形下,基于B样条的稳健回归算法比常规的最小二乘回归算法具有更好的表现。The fitting function model is constructed by making use of B-spline basis.Aiming at the outliers in the measuring data,robust function is made use of for robust regression instead of the ordinary least square regression and the fitting model is solved by iterative weighted method.The test results show that based on the B-spline,the robust regression has better performance than the ordinary least square regression against the outliers.

关 键 词：B样条 数据拟合 稳健回归 

分 类 号：O241.5[理学—计算数学]