一种新的抗差估计方法及运算

A new robust estimation method and its calculation

为了解决经典的抗差估计方法中,由于常数项的选取使观测值的属性发生误判,影响最终估计值和迭代次数的问题,以和极大似然估计为准则,推导出一种新的抗差估计权函数,并从理论上论证了其特性。通过与经典的抗差估计方法相比较表明,新的估计方法不但具有较强的抗差性,并且能够弥补经典抗差估计方法存在的缺陷。该方法对于经典抗差估计的研究具有一定的指导意义。

In classical estimation method, the selection of constant value may result in the misclassification of the property of observations, and affect the final estimation results and iteration times. Based on the theory of sum maximum likelihood estimation and standardized normal distribution, a new weight function of robust estimation is presented in this paper. The characteristics of this function are proven theoretically： In comparing with classical estimation method, it shows that the new method not only has the characteristics of a strong robust estimation method, but also can overcome the shortcomings of classical estimation method. The new method is also useful for studying classical estimation method.