摘要
本文首先介绍最小绝对误差和回归方法中的BR算法,然后根据线性规划中的灵敏度分析,对最小绝对误差和回归提出两种分别按样本点和阶进行递推的快速辨识算法,为最小绝对误差和回归的在线辨识和实时控制奠定基础。
This paper firstly introduces the most efficient Barrodale and Roberts' algorithm in all regressive algorithms of the minimum absolute errors sum regression. Then,based on Barrodale and Roberts' algorithm and sensitivity analysis in linear programming, two recursive algorithms of the minimum absolute errors sum regression are proposed and discussed in the paper.
出处
《控制与决策》
EI
CSCD
北大核心
1991年第5期345-350,共6页
Control and Decision
关键词
绝对误差
回归
递推
辨识
算法
regression, identification,recursion,sensitivity analysis
