Chemical imaging for determining the distributions of quality components during the piling fermentation of Pu-erh tea

Abstract

Piling fermentation (PF) is the key to the formation of Pu-erh tea quality; however, the traditional PF process limits the digital and intelligent production of Pu-erh tea. To establish qualitative and quantitative prediction models for the PF degree of Pu-erh tea, hyperspectral imaging technology and chemometric analysis were utilized. A qualitative model that uses least-squares support-vector-machine effectively distinguished the PF degree with an accuracy of 98.63%. Moreover, the chemical contents of quality-affecting components, namely total catechin, free amino acids, and chlorophyll a, were accurately quantified using raw spectral data with residual prediction deviations of 11.26, 4.34, and 3.89, respectively. The spatial distribution of these components during PF was mapped through chemical imaging, and the PF was deemed complete when the model predicted that the total catechin, amino acid and chlorophyll a content were less than 0.48, 11.21 and 1.29 mg/g, respectively. These findings provide a theoretical foundation for digital processing.