论文部分内容阅读
多变量分析能充分利用现代分析仪器获取的多通道量测数据,解决多组分不经分离或掩蔽同时测定的问题。然而,噪声的存在往往影响多变量分析的准确度。本文用仿真数据研究了噪声在不同分离度下,不同半峰宽比时和不同信噪比下,对3种多变量分析方法——卡尔曼滤波法,多元线性回归法和主成分回归法的影响。实验表明,噪声影响多变量分析方法的准确度,且噪声对主成分回归法的分析结果准确度影响较大,而对卡尔曼滤波法和多元线性回归法影响较小。
Multivariate analysis can make full use of multi-channel measurement data acquired by modern analytical instruments to solve the problem of simultaneous determination of multiple components without separation or masking. However, the presence of noise often affects the accuracy of multivariate analysis. In this paper, the simulation data of three different multivariate analysis methods - Kalman filtering, multivariate linear regression and principal component regression were used to study the noise under different resolutions, different half-width-width ratios and different SNRs influences. Experiments show that noise affects the accuracy of the multivariate analysis method, and the noise has a great influence on the accuracy of the analysis results of the principal component regression method, while has little effect on the Kalman filtering method and the multiple linear regression method.