节点文献
采用主成分分析结合相关系数法提高苯及其同系物含量预测结果的准确度
Adopting the Method of Principal Components Analysis Combined with Correlation Coefficient to Increase the Predicted Concentration’s Accuracy of Benzene and Its Homology Mixture
【摘要】 采用短波近红外光谱法对苯及其同系物进行了检测 ,研究了多组分近红外光谱主成分的特点和物理意义。指出前几个主成分和此多组分混合物的相关系数曲线十分相似 ,给予理论证明并指出了成立的条件。主成分结合相关系数法可以鉴别出系统高频噪声 ,扣除这部分噪音可有效地提高预测模型的预测准确度。
【Abstract】 The concentrations of benzene and its homology mixture were measured by near infrared spectra, and the emphasis was put on the character of the principal component and its physical significance. It is pointed out that the anterior principal components are very similar to the correlation coefficient of the multi-component solution and the theoretical proof for the right condition is given. The high frequency noise of the system can be checked out by principal component combined with the correlation coefficient. Removing the noise can greatly increase the accuracy of the prediction model.
【Key words】 The principal component; Noise; Partial least-squares regression; Correlation coefficient; Near infrared spectrum; Benzene; Homologue;
- 【文献出处】 光谱学与光谱分析 ,Spectroscopy and Spectral Analysis , 编辑部邮箱 ,2004年12期
- 【分类号】O625;O657.3
- 【被引频次】18
- 【下载频次】314