选权迭代定位粗差时权函数参数之功能

Effects of Parameters of Weight Function for Blunder Detection by the Recursive Weighted Least Squares Method

  • 摘要: 本文研究了幂函数型权函数的参数与定位粗差能力的关系。提出了权函数基本变量的幂取值范围为-2.5~-4.0;计算权的基本变量对于第1,2次迭代用"放大余差",以后则用余差;对余差进行统计判断的统计量采用"标准化余差"。

     

    Abstract: Based upon an analysis of the relationship between patrameters of the weight function of the negative power form and the capability of locating blunders, it is found necessery to handle separately the choice of the basic variable of the weight function and the test value for statistical tests. The result of investigation shows that the so-called "enlarged residual" should be used as the basic variable of the weight function for the first two iterations, and it is better to choose the residual itself for the following. The "Standardized residual" is used as the statistical test value, while the power value of the weight function is taken from - 2.5 to - 4.0.

     

/

返回文章
返回