Tea Quality Analysis By Boosting Combined With Near Infrared Spectroscopy

Although tea quality analysis based on the modern near infrared spectroscopic technique has achieved some progresses, attention on this may still need to be further paid. For instance, as to the quantitative analysis of tea, the influence of the sample quantity variation on spectra is always omitted, greatly reducing the measurement accuracy. Then, the determination of sample is still confined to a single composition. In addition, the chemometric method in exist may not mine the whole information behind the near infrared spectral data, and so on. In allusion to the defects on the analysis of the tea quality, the following two researches have been implemented in the current thesis:(1) The ytterbium-based internal standard near infrared spectroscopic measurement technique and multivariate calibration correction methods have been employed for tea quantitative analysis. Ytterbium (Yb), as a rare earth element, aimed to compensate for the variation of spectra induced by the sample quantity alteration during the spectral measurement on the basis of the powdered samples. The boosting technique has been introduced in the present study to enhance the performance of LS-SVR, forming a new type of multivariable modeling algorithm, the boosting least-squares support vector regression (BLS-SVR), used to implement the multivariable modeling task. The results showed that it gives satisfactory accurate results using the Yb-based internal standard NIR spectroscopy coupled with the BLS-SVR method, which can achieve accurate, rapid and low cost analysis of tea quality. Moreover, the introduction of boosting drastically enhances the performance of individual LS-SVR and BLS-SVR compares favorably with the conventional partial least-squares regression (PLSR).(2) Boosting has been combined with partial least-squares discriminant analysis (PLS-DA) to develop a new pattern recognition method called boosting partial least-squares discriminant analysis (BPLS-DA). BPLS-DA is implemented by firstly constructing a series of PLS-DA models and then combining the predictions from the constructed PLS-DA models to obtain the integrative results. Coupled with near infrared (NIR) spectroscopy, BPLS-DA has been applied to discriminate different kinds of tea varieties. As comparisons to BPLS-DA, the conventional principal component analysis (PCA), linear discriminant analysis (LDA), and PLS-DA have also been investigated. Experimental results have shown that the inter-variety difference can be accurately and rapidly distinguished via NIR spectroscopy coupled with BPLS-DA. Moreover, the introduction of boosting drastically enhances the performance of an individual PLS-DA, and BPLS-DA is a well-performed pattern recognition technique superior to LDA.