论文部分内容阅读
为了改善近红外(NIR)光谱测量精度,提出了以中等浓度样品作为参考物的测量方案。从理论上分析了扣除参考背景后吸光度的误差来源,发现以中等浓度样品作为参考物有可能减小样品池和光纤损耗导致的测量误差。以葡萄糖为分析对象,开展了葡萄糖水溶液和3%Intralipid仿体溶液的近红外透射验证实验。在糖水溶液的双光路实验中,以中等糖浓度样品作为参考物建立的偏最小二乘(PLS)模型的校正均方根误差(fRMSEC)和交互验证均方根误差(fRMSECV)比纯水背景分别下降了36.6%和41.0%;而仿体实验中,以中等糖浓度样品作为参考物时,模型的校正均方根误差和交互验证均方根误差比无糖3%Intralipid溶液作为参考物时分别下降了66.3%和39.6%。实验结果表明以中等浓度样品作为参考物能有效提高近红外透射光谱的分析精度。
In order to improve the NIR spectral measurement accuracy, a measurement scheme using a medium concentration sample as a reference is proposed. The source of error in absorbance after subtracting reference background was theoretically analyzed. It was found that it is possible to reduce the measurement error caused by sample cell and fiber loss by using medium concentration samples as reference materials. Using glucose as analyte, the near-infrared transmission experiment of glucose solution and 3% Intralipid solution was carried out. In the dual optical experiments of aqueous sugar solution, the corrected root mean squared error (fRMSEC) and the cross-validated root mean square error (fRMSECV) of the partial least squares (PLS) model established with the medium sugar concentration samples as the reference were higher than that of the pure water background Respectively, decreased by 36.6% and 41.0%, respectively. However, when the samples were used as reference material, the root mean square error (RMSE) and root mean square error (RMSE) of the model were higher than those of the sugar-free 3% Intralipid solution Decreased by 66.3% and 39.6% respectively. Experimental results show that the medium concentration of samples as a reference can effectively improve the analytical accuracy of NIR transmission spectroscopy.