烟草中水溶性灰分碱度的近红外分析

Analysis of the Alkalinity of the Water-soluble Ash in Tobacco with Near-infrared Method

[目的]应用偏最小二乘法建立烟草水溶性灰分碱度的近红外分析模型。[方法]采集标样的傅里叶变换近红外漫反射光谱数据并利用化学计量学的方法对所得到的近红外光谱数据与灰分碱度的基础数据进行建模,优化建立后的模型,并对该模型进行验证。[结论]经模型优化后最终的相关系数为0.978 0,均方估计残差为0.040 0,预测标准差为0.245,交互验证均方差为0.089 3,模型优化指数为74.1,主因子为11。用57个验证样品对模型进行检验,结果显示:平均误差为0.193 4,预测范围为0.15～3.94。[结论]该方法具有较好的可行性。

[Objective] The model of the near-infrared analysis for water-soluble ash alkalinity with the method of partial least squares was established.[Method] The data,collected with near infrared spectroscopy and then analyzed with chemometric method,of the standard tobacco sample was treated with the basic data of its ash alkalinity for the establishment of model.The model was tested its optimizing.[Results] The final correlation coefficient of the optimized model was 0.978 0,which residual difference of estimated mean square was 0.040;forecast standard deviation,0.245;mean square deviation of cross-validation,0.089 3;optimization index of model,74.1 and main factor,11.The average error of 57 validated sample from the result-testing of the model was 0.1934 with the predicted range of 0.15-3.94.[Conclusion] The method was with good feasibility.