一种改进的一元线性回归算法  

A New Unitary Linearity Regression Algorithm Constringed by Specified Error

作　　者：华德宏[1] 刘刚[1] 兰家隆[1] 

机构地区：[1]电子科技大学电子工程学院,四川成都610054

出　　处：《现代电子技术》2006年第7期63-65,68,共4页Modern Electronics Technique

基　　金：十五国防预研基金资助项目(41101030301)

摘　　要：提出了一种改进的一元线性回归算法。该算法基于递推一元线性回归算法,通过收敛判据的设定能自动收敛到给定误差,收敛速度提高到未加判据时的4倍。仿真结果显示,该算法具有很好的收敛性能,计算量和存储量都大幅度下降,具备良好的工程应用价值,已在数字信号处理器上得到了硬件验证。A new unitary linearity regression algorithm constringed by specified error is proposed in the paper. The new algorithm is constringed to specified error automatically by criterion of convergence designed, the time of computing reduces by one fourth because of the criterion of convergence. The computer simulations show very good convergence ability,the computations and the memory reduce a lot, the new algorithm has great value of engineering applications which has been validated on the digital signal processor.

关 键 词：线性回归 最小二乘算法 卡尔曼滤波 信号处理 

分 类 号：TN911.7[电子电信—通信与信息系统]