摘要
在估计谐波分布的特征参数时,一般将非线性谐波分布函数转换为线性函数,然后用线性最小二乘法处理数据。这样求解会受到异常点干扰,并可能导致有偏估计,增大标准差。本文提出优化理论为基础的估计谐波分布参数的一种数值方法。谐波实验研究证明这种方法可以有效地衰减数据异常点的扰动误差,比较真实地反映测量数据的固有特性。
In estimating the harmonic distribution parameters, the method of linear leastsquares is often used as the data processing after transforming the harmonic distribution function from nonlinear to linear. The solution using this method can be disturbed by the outlier of measuring data and is likely to produce a biased estimation and large standard deviation. Therefore, a numerical technique using optimization theory to characterize the parameters of the harmonic distribution is proposed. This method can avoid the bug of leastsquares method, deaden efficiently the disturbing errors of the outlier of measuring data and reflect truly the nature identity of measuring data.
出处
《河南科技大学学报（自然科学版）》
CAS
2003年第3期28-29,共2页
Journal of Henan University of Science And Technology:Natural Science
基金
国家自然科学基金资助项目(59805007)
河南省杰出青年基金资助项目(02120001500)
河南省自然科学基金资助项目(0311061400)
