低含水率对烟样化学成分近红外测定的影响
Effects of low moisture contents on near-infrared determination of chemical composition for tobacco samples
-
摘要:
为探究样品的含水率对近红外预测结果的影响，利用预先建立的近红外模型对含水率低于建模样本集的复烤片烟粉末样品常规6项化学成分（烟碱、总糖、还原糖、总氮、钾和氯含量）进行预测。结果表明，低含水率样品的烟碱、还原糖和总氮含量的近红外预测值小于化检值，而总糖、钾和氯含量的预测值明显大于化检值。低含水率的样品经回潮后含水率符合建模样品集的含水率范围，模型对样品的6项化学成分的预测值接近化检值，说明应用近红外测定复烤片烟粉末样品的化学成分时，测试样品的含水率和建模样品集的含水率范围相符是确保模型预测准确性的重要条件。 Abstract:To explore the effects of sample moisture content on the near-infrared prediction results, using pre-established near-infrared models, contents of six chemical components （nicotine, total sugar, reducing sugar, total nitrogen, K and Cl） of redried strip powder samples with moisture content lower than that of the samples used to establish the models were predicted. The results indicated that predicted values of contents of nicotine, reducing sugar and total nitrogen were lower than the test values, while the predicted values of contents of total sugar, K and Cl were higher than the test values. After conditioning, the range of moisture contents of these low moisture content samples was within the moisture content range of the model sample set and the predicted values of the six chemical components were close to the test values, which indicated that test samples with moisture contents within the moisture content range of the model sample set were an important condition for the accuracy of model prediction.
下载: