补偿波段建模方法及其在血清葡萄糖光谱分析中的应用  

The compensation model based on continuous waveband for the human serum glucose applied through the FTIR/ATR spectrum

作　　者：谢军[1,2] 马辉[1] 潘涛[3] 

机构地区：[1]海南大学机电工程学院,海口570228 [2]清华大学深圳研究生院,深圳518055 [3]暨南大学理工学院光电工程系,广州510632

出　　处：《分析试验室》2015年第3期255-258,共4页Chinese Journal of Analysis Laboratory

基　　金：国家自然科学基金项目(61078040;11374179)资助

摘　　要：利用傅立叶变换红外光谱（FTIR）和衰减全反射（ATR）技术,建立血清葡萄糖的快速定量分析方法。基于模型预测效果,提出了一种联合最佳单波段的补偿波段建模方法。在全谱（4000~600 cm-1）区间,根据移动窗口偏最小二乘法得到人体血清葡萄糖的最佳单个连续波段后,另外选择一个波段进行补偿建模,优选出最佳补偿波段。结果表明,这种补偿波段建模方法建模效果不仅仅远好于全谱,也好于最佳单个波段。用独立的实际样品对模型进行测试,补偿模型的预测均方根偏差（RMSEP）和预测相关系数（Rp）分别为0.782 mmol/L和0.944。The rapid quantification method of human serum glucose was established by using the fourier transform infrared spectroscopy（ FTIR） and attenuated total reflection（ ATR）. Based on the best predicting effect,a compensation band model combining with best single continuous waveband was developed. Moving window partial least squares（ MWPLS） of whole spectrum（ 4000 ~ 600 cm-1） was carried and the best single continuous waveband was found. Then an appropriate compensation waveband was selected combining with the best single waveband. The results show that the prediction bias of the compensation model for human serum glucose was obviously better than that of the whole spectral, and was better than MWPLS model also. Finally, the independent test set was tested for model verification. RMSEP and RPwere 0. 782 mmol / L and 0. 944 respectively by the compensation model.

关 键 词：FTIR ATR 血糖 波段选择 移动窗口偏最小二乘 补偿模型 

分 类 号：O657.3[理学—分析化学]