近红外技术应用于成品卷烟的内在质量稳定性控制

Application of the NIR Spectroscopy to the Control of the Qualitative Stability of the Manufactured Cigarette

以某品牌卷烟为例,采用傅立叶变换近红外光谱技术采集成品卷烟的近红外光谱,结合连续流动法测定样品中还原糖、烟碱、氯、钾含量的数据,采用偏最小二乘法(PLS)分别建立了预测品牌卷烟烟丝中还原糖、烟碱、氯、钾及糖碱比、钾氯比的近红外模型,模型的相关系数(除钾外)都在0.95以上。用建立的模型快速检测红狼卷烟的各化学成分,通过值的波动范围初步监控成品的质量稳定性。并采集它们的近红外光谱,应用连续小波变换(CWT)对原始光谱进行预处理,再对CWT系数进行主成分分析(PCA),在PCA分析的基础上引入马氏距离算法建立成品卷烟质量稳定性评价模型。通过研究光谱预处理对卷烟样品分类鉴别的影响,确定马氏距离阈值Dthro,结合近红外定量测定值与马氏距离阈值Dthro,对卷烟产品质量稳定性进行评价和控制。

Taking a certain brand of cigarettes as an exanlple, which applied Fourier transform near-infrared Spectrum spectroscopy to acquire near-infrared spectroscopy of manufactured cigarettes, combines with autoanalyzer to determine the data of reducing sugar, nicotine, chlorine, potassium in the sample, and uses PLS to establish the near-infrared model to predict the ingredients of the brand tobacco： reducing sugar, nicotine, the ratio between them,chlorine, potassium and the ratio between them, getting the correlation coefficient （in addition to K） above 0.95. The chemical composition of SEFTWOLVES cigarette can be detected rapidly with the help of the establishment of the model, and the qualitative stability of the manuffactured cigarettes can be monitored by fluctuations of/he results. At the same time,in order to establish the model of quality stabihty of the manufactured cigarette evaluation, it need the following steps： first collected SEPTWOLVES cigarettes, then acquired their near-infrared spectroscopy, next applied continuous wavelet transform （CWT） to pre-process the original spectrum, and carried out principal component analysis （PCA） to analyze CWT coefficients, on the basis of the analysis, Mahalanobis distance Algorithm can be introduced. Determining the threshold of Mahalanobis distance by studying the impact of pre-processing performance on the classification of cigarette samples, and combining near-infrared quantitative measurement value with Mahalanobis distance threshold to evaluate and control the quality stability of the cigarette.