半参数模型估计的补偿最小二乘法与正则核方法

Penalized least squares method and the kernel method of Semi-parametric model estimation

本文根据测量数据处理中利用半参数对系统误差去除的效果,提出了半参数补偿最小二乘法和正则核方法两种方法进行比较。两种去除系统误差的方法都存在着局限性,应该根据系统误差的具体情况,选择不同的方法;最后通过模拟算例去除系统误差,并与实测数据进行拟合验证两种方法的可行性。

Based on the effect of systematic error removal using semi-parameters in measuring data processing,the paper compared between semi-parametric penalized least squares method and the kernel method.Because both methods exists limits,different methods should be selected depending on practical cases.The paper finally gave simulation examples to remove systematic errors,and fit with actual measurement to validate the feasibility of two methods.