摘要
在近红外分析中，样品状态与装样条件对光谱影响较大。本文对茶样制备的颗粒度、厚度、松紧度进行了研究，以吸光度重复性STD值和差谱技术DH值为评价指标，结合正交试验及方差分析确定最佳制样条件，并利用PLS建立模型进行验证。结果表明，样品粒度为40目-60目，压样力度40MPa，厚度为4mm时谱图重合性最佳；以最佳制样条件制各样品并建立红茶全氮量和咖啡碱预测模型，预测相关系数分别为0．9666、0．9767；预测均方根误差RMSEP分别为O．0266、0．029。茶叶样品近红外光谱检测模型的预测精度大大提高。
As a powerful analytical tool in product quality determination, NIR spectra are easily affected by sample＇s state and loading condition. To provide foundation with optimum test condition when modeling, the influence of sample size, sample thickness and sample tightness were studied. Using orthogonal experimental design, the best sample preparation conditions were determined by the values of photometric precision （STD） and subtractive spectroscopy （DH） . Experiments indicated that the repeatability of spectral was best with the powder＇s particle size of 40--60mesh, 40MPa pressure sample strength and 4-millimetre thickness. Based on near infrared spectroscopy technique and partial least squares （PLS） , models of the total nitrogen and caffeine content were established .Two models were developed under the best sample preparation conditions. The predicting correlation coefficient and the root mean square error of prediction （RMSEP） of total nitrogen and caffeine in black tea were respectively 0.9666, 0.9767 and 0.0266, 0.029. The near infrared spectrum models of tea sample detection precision was improved.
出处
《光谱实验室》
CAS
2014年第2期238-245,共8页
Chinese Journal of Spectroscopy Laboratory
基金
安徽省高校省级科学研究重要项目（KJ2013A109）：国家星火计划重大项目（2012GA710001）
关键词
近红外光谱
样品制备
模型优化
Near Infrared Spectroscopy
Sample Preparation
Model Optimization
