Non-destructive Detection of Vitamin C, Sugar Content and Total Acidity of Red Globe Grape Based on Near-Infrared Spectroscopy

The sugar content and total acidity of red globe grape directly affect the taste and quality of fresh food taste and its by-products. Vitamin C is a necessary nutrient for human beings, and it has become the main index to evaluate the quality of red globe grape. The traditional detection method of red globe grape internal quality is destructive sampling, which is cumbersome and time consuming, and has many drawbacks. In this work , based on near-infrared spectroscopy, the rapid nondestructive detection of red globe grape Vc, sugar content and total acidity was performed. The spectral data of red globe grape samples were collected, and the competitive adaptive reweighed algorithm , stability competitive adaptive reweighed sampling algorithm and successive projection algorithm were respectively applied to extract an effective characteristic band. Then the content of Vc and sugar and the total acidity were measured comparatively , and a corresponding partial least squares regression model was established. SPA was combined to extract a secondary characteristic band on the extraction of an effective characteristic band, and a corresponding PLSR model was established. The results showed that the correlation coefficient between the correction set and the prediction set of the PLSR model established by the secondary characteristic band was higher than that established by the primary characteristic band extraction, and the root mean square error of the model was reduced. The correlation coefficients of correction set and prediction set of the optimal PLSR model for red globe grape Vc, sugar content and total acidity based on the optimal band points extracted from the secondary characteristic band were 0. 983 , 0. 982 and 0. 976, respectively, and the correlation coefficients of prediction set were 0. 975, 0. 980 and 0. 975, respectively. This stable model built with fewer bands predicts Vc, sugar content and total acidity, and greatly reduces run time. This model provided a technical support for subsequent portable detector and online dynamic detection research.