Non-Destructive Quantitative Analysis of Grape Components Based on Visible–Near-Infrared/Shortwave–Near-Infrared Spectroscopy and Chemometrics

The contents of sugar and acid components in grapes are important indicators for evaluating their quality and maturity. To achieve rapid, non-destructive, and accurate detection of glucose, fructose, tartaric acid, and malic acid in grapes, this study selected 60 mature grape varieties as research objects. Spectral data were acquired using a visible–near-infrared/shortwave–near-infrared (VNIR/SWIR) hyperspectral system, and the contents of the four components were determined by high-performance liquid chromatography (HPLC) as reference values for modeling. Based on spectral preprocessing and CARS-based feature-variable selection, PLSR, SVM, and BPNN prediction models were constructed using full variables and selected feature variables, respectively, and their performance was evaluated using R2, RMSE, and RPD. The results showed that the feature-variable models outperformed the full-variable models, with the NIR/SWIR-CARS-SVM model performing best, with RP2 values reaching 0.8630, 0.8441, 0.8546, and 0.8283; RPDP values reaching 2.7287, 2.5782, 2.6451, and 2.4145; and RMSEP values as low as 3.5114, 3.4751, 0.5310, and 0.1646 for glucose, fructose, tartaric acid, and malic acid, respectively. This study provides an effective method for the non-destructive quantitative detection of sugar and acid components in grapes. Under the single-site, single-season conditions of this trial, the 60 varieties differed substantially in glucose, fructose, tartaric acid, and malic acid. These findings suggest variety-level differences that warrant further validation across multiple vintages and geographical regions before industrial recommendations can be made.