摘要
在数据拟合中 ,当个别数据存在较大测量误差时 ,采用 L1 准则对数据进行拟合要优于最小二乘准则。本文通过建立 L1 准则与最小二乘准则之间的关系 ,给出了一种基于 Gauss- Newton法的L1 数据拟合算法。计算结果表明 ,该方法易于理解 ,便于编程 ,具有一定的理论意义和现实意义。
It is a good choose to apply L\-1 Criterion other than least square criterion in data fitting when one or two large errors exist in the measuring data.This paper introduces a new algorithm for L\-1 data fitting based on Gauss\|Newton method through establishing the connection between L\-1 Criterion and least square criterion.Experimental results show that the method can be realised easily.
出处
《航空计测技术》
2000年第5期11-12,26,共3页
Aviation Metrology & Measurement Technology
