The invention relates to the technical field of
food quality detection, in particular to a
green tea variety and producing area identification method based on multi-
solvent fusion
surface enhanced Raman spectroscopy, which comprises the following steps: crushing and sieving a
green tea sample, adding
ultrapure water, absolute ethyl
alcohol or
hydrochloric acid solution, oscillating and extracting, centrifuging, and reserving supernate; mixing the supernate with the
silver nanoparticle suspension, performing ultrasonic treatment, dripping on a
silicon wafer,
drying, acquiring spectrums at three positions by adopting a
laser confocal microscopic Raman
spectrometer, and performing baseline correction to obtain an average spectrum, thereby obtaining a surface enhanced
Raman scattering spectrum; carrying out data fusion on the surface enhanced
Raman scattering spectrum through matrix splicing, CARS, PCA, t-SNE and UMAP; according to the method, the fusion spectrum is divided into a
training set and a
test set, an identification model is constructed by adopting SVM, KNN and RF,
green tea variety classification and origin
traceability are realized, spectrum data are processed from the aspects of complete information retention,
feature screening, linear
dimensionality reduction,
local structure retention and multi-
scale structure retention, and the classification accuracy reaches 98.21%.