摘要
针对测量数据处理中系统误差对参数估计的影响,提出了半参数估计的正则核方法。这种方法可以在不知道系统误差具体形式的情况下有效地分离系统误差的影响,并且对系统误差进行估计。由于正则项的引入,有效地避免了对非参数部分的过拟合现象,同时对于小样本的观测值也能取得较好效果。最后通过两个算例证明了该方法的有效性。
An estimation method of semi-parametric models based on regular kernel function was presented. It could effectively separate and estimate systemic error, aiming at the influence of systemic error for estimating parameter in the cause of surveying data processing. Furthermore, as a result of affiliating the regular parameter, the method could avoid excessively fitting nonparameter and obtain satisfying effect when dealed with small observed sample values. Finally with two examples the validity of the estimations was verified.
出处
《测绘科学技术学报》
北大核心
2008年第4期284-287,共4页
Journal of Geomatics Science and Technology
基金
国家自然科学基金项目(40474007)
关键词
半参数估计
正则化
核函数
系统误差
semi-parametric models estimation
regularization
kernel function
systemic error
