近红外光谱分析定标技术介绍实验室分析.ppt

* * * 标准偏差 根据正态（高斯）分布，有 68% 的结果落在此误差范围内 误差分布 样品分析误差符合高斯分布 (Gaussian) : 100 样品 ，验证样品平均浓度为4%, CV=1%: 32 误差落在 ±0.04% 5 误差落在 ±0.08% 1 误差超过 ±0.12% 3.8 3.9 4.0 4.1 4.2 68% 95% 99% Slope斜率 k 为斜率， m 为截距 Bias系统偏差 WIN ISI II - Scatter plot Correlation 相关系数 表征x,y之间的相关关系 如果x,y共变化，则可获得好的相关 Correlation相关系数 1-VR = 1-SECV2/SD2 近红外技术应用简述 (1) *检查仪器是否在正常的工作状态 *收集样品并获得样品的近红外光谱 *对样品进行精确的实验室分析并将分析结果输入到光谱文件 *确定“最好的”样品 ... *利用样品的光谱数据及化学数据建立数学模型来预测样品的成分 *验正模型 (定标方程)的准确性 *将模型(定标方程)应用于常规分析 近红外技术应用简述 (2) ... *对模型(定标方程)进行周期性的验正 *如需要，应对模型 (定标方程)进行升级 近红外技术应用简述 (3) * * * * * There are two abbreviations depending of how the NIR light is applied: NIR for reflectance and NIT for transmittance. Sometimes you can also see the abbreviations NIRT (Near Infrared Transmittance) and NIRR (Near Infrared Reflectance) * This picture illustrates the difference between transmittance and reflectance. It is important to point out that with transmittance in the NIR range it is possible to penetrate up to 30 mm of the sample and the entire path length of the sample is integrated into the spectral measurement. With reflectance only 0.1-1 mm of the surface will be penetrated which will craete problems for non-homogenous samples. * In the first step of the evaluation, 10 or more samples are split into blind duplicates and encoded with new sample numbers. It is very important that the sample split be done as equally as possible and that the original identity of the samples be hidden during the subsequent analysis. The samples are then analyzed according to the Standard Operating Procedures in the laboratory. Maybe do this as a seperate hand-out!!! * In the first step of the evaluation, 10 or more samples are split into blind duplicates and encoded with new sample numbers. It is very important that the sample split be done as equally as possible and that the original identity of the samples be hidden during the subsequent analysis. The samples are then analyzed according to the Standard Operating Proc