In situ monitoring of total polyphenols content during tea extract oxidation using a portable spectroscopy system with variables selection algorithms
This work presents the rapid monitoring of the total polyphenols content during tea extract oxidation using a portable spectroscopy system. Firstly, an in situ monitoring installation for tea extract oxidation was developed, including a tea extract oxidation system and a spectroscopy system for spectra acquisition. Then, partial least squares (PLS) regression with several variables selection algorithms was used for modeling. Synergy interval partial least square (Si-PLS), genetic algorithm (GA), competitive adaptive reweighted sampling (CARS) and ant colony optimization (ACO) algorithms were used comparatively for selecting the most effective variables. The performance of the final model was evaluated according to the correlation coefficient (Rp) in the prediction set. Experimental results showed that the variables selection methods could significantly decrease the number of variables and improve the model performance, especially for the ACO algorithm with the least variables. Finally, 28 independent samples were used to test the performance of the spectroscopy system, and the coefficient of variation (CV) of the final results was used to state the stability and reliability of this system. Results also showed that the CVs for most of the samples were less than 10%. This study demonstrated that the tea extract oxidation system combined with a spectroscopy system is a promising tool that could be used for in situ monitoring of tea fermentation.