Non-destructive prediction of tea polyphenols during Pu-erh tea fermentation using NIR coupled with chemometrics methods

The degree of fermentation is crucial for the tea quality. As fermentation duration increases, tea polyphenol content decreases. Exploring total polyphenol content is conducive to achieve the most appropriate fermentation degree. The optimum fermentation of Pu-erh tea liquor was determined in this study by employing the response surface methodology. The response surface experiments were designed with pH, inoculation amount, and solid-liquid ratio as the investigating factors and tea polyphenol degradation rate as the response variable. Under optimized conditions, raw NIR spectra were collected from Pu-erh tea fermented liquor, and standard normal variables transformation (SNV) was applied to eliminate noise interference. Afterwards, three variable screening methods were comparatively applied to select important variables. In comparison, the SNV-competitive adaptive weighted sampling-partial least squire (SNV-CARS-PLS) model achieved the best results by a total of 52 variables were selected from SNV preprocessed NIR spectrum for total polyphenol content in Pu-erh tea liquor, with a correlation coefficient of prediction (Rp = 0.9088), root-mean-square error of prediction (RMSEP = 0.0636 mg/g), and residual prediction deviation (RPD = 2.372). The developed method achieved a limit of detection (LOD) of 0.1908 mg/g and validation outcomes by standard method were satisfactory (p > 0.05) indicating that the developed method could be applied for the determination of tea polyphenols content, thus providing theoretical guidance for the establishment of an intelligent system to monitor the fermentation process of Pu-erh tea in real-time.