The Uncertainty Evaluation On Mid-infrared Spectra For Tea Examination

With the steady growth of China’s tea production and exportation, the requirement for the quality and security level of tea has been constantly raised. The evaluation of tea quality relies heavily on the sense of tea tasters, which may severely compromise the objectivity of the evaluation. In this paper, an experimental curve uncertainty model is used to study the effects of tea sample preparation condition on the measurement errors of IR spectrum of tea.The traditional uncertainty model applies only to the statistical interpretation of repeated measuring error of individual experimental data, but it does not apply to the IR spectrum experimental curve. In this paper, the traditional uncertainty model was expanded to establish an uncertainty model of infrared absorption spectrum curve of tea, which is used to measure and compare the infrared absorption spectrum of tea samples of different granularity. With this method, we draw the conclusion that the smaller the value of uncertainty, the better it is for the repeatability of infrared absorption spectrum of tea, and the larger the correlation coefficient of the measurements of spectrum curve. The result verifies the correctness of experimental curve uncertainty model we established.The experimental curve uncertainty model is then used to measure the infrared absorption spectrum of tea, and the effects of tea sample preparation condition on the measurement errors of IR spectrum of tea are studied. We analyzed the effecting law of five sample preparation condition parameters, which include tea powder granularity, the mixing ratio of tea and KBr, mixture grinding time, total mass of samples as well as squashing pressure and squashing time, on the measurement error of IR spectrum, and decided the best sample preparation condition of IR spectrum of tea as follows: tea powder granularity:240mesh; the mixing ratio of tea and KBr:1:120; mixture grinding time:5minutes; total mass of samples:120mg; squashing pressure:18Mpa; squashing time:3minutes. Verifying the feasibility of analyzing spectrum curve on uncertainty definition, we combine principal component analysis with uncertainty calculation. Calculation reveals that the law of data change derived from this method is the same as the previous one, but its extent of uncertainty is larger than that of average absorbance method, which can redound to the chosen of standard sample preparation condition, and this also provides a new approach for similar experimental analysis in the later work.