基于多元线性回归的雷达在线测流系统研究应用

Research and Application of RG-30 Based on Multiple Linear Regression

为提高雷达在线测流系统的测验精度,通过拉伊达准则剔除粗大误差,以相对误差平方和最小为目标函数,利用多元线性回归分析方法,建立了多探头水面流速与断面平均流速的多元一次拟合公式。通过t检验,分量级、分类型探讨了雷达在线测流系统探头安装位置与最优数目问题。最后,以白家川站和丁家沟站为例,比较了几种雷达在线测流系统系数率定方法。结果表明,以相对误差平方和最小为目标函数的多元线性回归方法提高了拟合精度,降低了建设维护成本。

In order to improve the measuring accuracy of RG-30,the gross error was canceled by Rayda's criterion.Taking the minimum relative error’s sum of squares as the target function,this paper established the multiple linear fitting formula by apply⁃ing multiple linear regression.,and discussed the problem of RG-30’s probe installation position and optimal number on different discharge stages and categories by t test.At last,several methods of RG-30 coefficient calibration were compared at the Baijiach⁃uan and Dingjiagou gauging stations.The results show that the method of multiple linear regression can improve the measuring ac⁃curacy of RG-30.Besides,it can reduce the construction maintenance costs.