基于近红外光谱技术的抹茶掺伪判别研究

发布时间：2018-07-02 23:58

本文选题：近红外光谱技术 + 抹茶　；参考：《安徽农业大学》2017年硕士论文

【摘要】：抹茶是茶叶深加工产品之一,因其良好的品质特点而展现出良好的经济效益。这也导致了目前市场上抹茶产品良莠不齐,出现了往抹茶中添加非茶成分,或以次充好的现象。这不仅损害了消费者的合法权益,也破坏了抹茶的市场秩序,为此建立快速可靠的抹茶掺伪检测方法对维护抹茶的市场秩序和品质安全具有重要意义。本研究主要针对市场上常见的抹茶掺伪物(白砂糖、麦芽糊精、桑叶粉、大麦苗粉)为研究对象,以近红外光谱技术为基础,并结合化学计量学方法(主成分分析结合线性判别分析、K最近邻法、偏最小二乘法),分别建立了纯抹茶与掺伪抹茶、4种掺伪抹茶的定性判别模型及其定量分析模型。(1)抹茶样品的前处理对定量模型结果的影响。在对抹茶进行压饼和未压饼的定量模型中,以添加白砂糖、麦芽糊精的抹茶为研究对象,压饼和未压饼的定量判别模型结果表明,未压饼的模型结果优于压饼的。这对于超微粉的近红外光谱技术分析有借鉴意义。(2)采用近红外光谱技术结合主成分分析和线性判别分析(PCA-LDA)、K最近邻法,对采集的42条纯抹茶样品与150条掺伪抹茶样品光谱(30条掺伪白砂糖抹茶光谱、30条掺伪麦芽糊精抹茶光谱、40条掺伪桑叶粉抹茶光谱、50条掺伪大麦苗粉抹茶光谱)建立定性判别分析来判别抹茶是否掺伪和掺伪类型,通过比较,PCA-LDA的结果优于K最近邻法。纯抹茶与掺伪抹茶、纯抹茶与掺伪白砂糖抹茶、纯抹茶与掺麦芽糊精抹茶、纯抹茶与掺桑叶粉抹茶、纯抹茶与掺大麦苗粉抹茶以及四种掺伪抹茶的定性分析模型的校正集识别率为98.3%、100%、91.7%、100%、100%,100%;预测集识别率96.5%、100%、87.5%、95.8%、90.3%、95.3%。由此可知,通过PCA-LDA建立的定性判别模型准确度和稳定性都很好,能够快速、准确的对抹茶中是否掺伪进行定性判别。(3)近红外光谱技术结合偏最小二乘法,使用四种光谱预处理方法对光谱预处理,并建立抹茶中4种掺伪物含量的定量模型,对添加白砂糖、桑叶粉、0~50%大麦苗粉、50~100%大麦苗粉的抹茶的光谱预处理方法以平滑处理的结果最好,麦芽糊精的以最大最小归一化预处理结果最好,所建立定量模型的校正集相关系数(Rc)分别为0.9910、0.9984、0.9975、0.9976、0.9975,RMSECV分别为0.391、1.60、1.11、2.44、0.66;预测集相关系数(Rv)分别为0.9992、0.9984、0.9976、0.9925、0.9977,RMSEP分别为0.365、1.99、1.13、1.93、0.761。由此可知,所建立的定量分析模型能够对抹茶中掺伪物的含量进行定量分析,且模型的准确度和稳定性能够满足一般的检测需求。
[Abstract]:Matcha is one of the deep processing products of tea, which shows good economic benefits because of its good quality. This also led to the current market for tea products mixed, the emergence of tea to add non-tea ingredients, or substandard phenomenon. This not only damages the legitimate rights and interests of consumers, but also destroys the market order of matcha. Therefore, it is of great significance to establish a fast and reliable method to detect the adulteration of matcha in order to maintain the market order and quality safety of matcha. In this study, the common adulterated matcha products (white sugar, maltodextrin, mulberry leaf powder, wheat seedling powder) were studied, based on near infrared spectroscopy (NIR). Combined with chemometrics (principal component analysis and linear discriminant analysis), (1) the effects of pretreatment on the quantitative model results were analyzed. (1) the qualitative discriminant model and the quantitative analysis model of four kinds of tea adulterated with pure and fake matcha were established, respectively. (1) the effect of pretreatment on the results of the quantitative model. In the quantitative model of pressing cake and unpressed cake of matcha, taking the tea with white granulated sugar and maltodextrin as the research object, the quantitative discriminant model of pressed cake and unpressed cake shows that the model of unpressed cake is better than that of cake. It is useful for the analysis of ultrafine powder by near infrared spectroscopy. (2) the near infrared spectroscopy combined with principal component analysis and linear discriminant analysis (PCA-LDA) is used. The spectrum of 42 samples of pure matcha and 150 samples of adulterated matcha (30 spectrum of adulterated white granulated tea and 30 spectrum of adulterated maltodextrin) were established. Qualitative discriminant analysis is used to determine whether or not matcha is adulterated and the type of adulteration. The results of PCA-LDA are better than that of K-nearest neighbor method. Pure matcha and adulterated matcha, pure matcha and adulterated white granulated tea, pure matcha and maltodextrin matcha, pure matcha and mulberry leaf powder matcha, The correct set recognition rate of qualitative analysis models of pure matcha, adulterated wheat seed powder and four kinds of adulterated matcha were 98.3 / 100 and 91.7 / 100 and 100 / 100, respectively, and the predictive recognition rate was 96.57.595 / 90.33 / 100 and 90.33 / 90.3 / 100, respectively. It can be seen that the qualitative discriminant model established by PCA-LDA has good accuracy and stability, and can quickly and accurately determine whether or not adulterated tea is adulterated. (3) Near-infrared spectroscopy combined with partial least square method. Four spectral pretreatment methods were used to pretreat the spectrum, and a quantitative model of the content of four adulterated compounds in mash tea was established. The spectral pretreatment of mash with 50% wheat seedling powder of 50% mulberry leaf powder and 100% wheat seedling powder with smooth treatment was the best, and that of maltodextrin with maximum and minimum normalization was the best. The calibration set correlation coefficient (RC) of the established quantitative model was 0.9910 / 0.9984 / 0.99755 / 0.99765 / 0.9975 / 0.9975 / 0. The RMSECV was 0.391U 1.601.111.111.44 / 0.66, respectively, and the correlation coefficient (Rv) of the prediction set was 0.99920.99840.99764 / 0.99725 / 0.9977N RMSEP was 0.3651.9ng1.131.93n / 0.761respectively, and the correlation coefficient (RV) of the prediction set was 0.99920.9984U 0.9976U (0.99725) and 0.99775 (0.99775), respectively. It can be seen that the established quantitative analysis model can quantitatively analyze the content of adulterate in tea, and the accuracy and stability of the model can meet the general needs of detection.
【学位授予单位】：安徽农业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O657.33;TS272.7

【参考文献】