非线性模型的最小最大绝对残差估计  

Estimation of minimal maximal absolute residual of non-linear model

作　　者：彭军还[1] 

机构地区：[1]桂林工学院土木工程系

出　　处：《桂林工学院学报》2002年第3期218-220,共3页Journal of Guilin University of Technology

基　　金：广西自然科学基金项目资助 (9912 0 0 8)

摘　　要：非线性模型的计算 ,首先对参数赋予初始值 ,最重要的是保证计算收敛 ,最小二乘法给残差以等权 ,对初值敏感 .最小最大绝对残差方法能保证最大绝对残差下降 ,这是非线性模型参数估计计算收敛的重要原因 ,算例表明最小二乘法比最小最大绝对残差方法对初值敏感 .In computing non-linear model, the estimated parameters are first initialized.It is essential to guarantee the coverage of computation. The least squares method gives all the residuals an equal weight which is sensitive to initial value. Minimizing the maximal absolute residual can guarantee the decreasing of the maximal absolute residual in computation which is the main cause to guarantee converage of non linear model computation. The example shows that the least squares method is more sensitive to the initial value of parameters than the minimal maximal absolute residual method.

关 键 词：非线性模型 最小最大绝对残差方法 最小二乘法 参数估计 测绘科学 

分 类 号：P207.1[天文地球—测绘科学与技术]