Apples are one of the four major fruits in the world. my country is the world's largest apple producer. The total production in 2006 was about 25.3 million tons, but the annual export volume only accounted for 1.5 of the total production. About %, which is extremely disproportionate to Apple’s status as the largest producer of apples. The main factors restricting my country's apple exports are poor fruit quality, low industrialization level, backward testing technology and evaluation standards that are not in line with international standards. At present, browning of apple pulp occurs very seriously during storage. Browning detection has shortcomings such as sample destruction, complex operation, long time consumption, high cost, and inability to achieve online detection. Therefore, based on studying the browning mechanism of apple pulp, exploring a method for rapid non-destructive detection of pulp browning based on near-infrared spectroscopy, and establishing a mathematical model with stable performance is a task of great theoretical significance and practical application value. This article takes Red Fuji apples as the research object, and takes the mechanism of browning and non-destructive testing technology as the research goals. It studies the browning mechanism of apple flesh, near-infrared spectrum response characteristics, influencing factors of browning degree non-destructive testing, and near-infrared spectroscopy detection. Systematic research was conducted on matching parameters and other aspects, and on this basis, a rapid non-destructive detection model for apple pulp browning was established. The main results of the study are as follows: (1) Through the study of the browning mechanism of apple pulp, the occurrence of enzymatic browning, Maillard reaction and ascorbic acid oxidation reaction in apples during different storage periods was comprehensively analyzed. The results showed that the mechanisms of browning of apple pulp are different at different storage stages. In the early stage of storage (before 80 days), the main mechanism of browning is the Maillard reaction of sugars and amino acids. In the late stage of storage (after 80 days), The main mechanism of browning is enzymatic browning catalyzed by PPO and POD. The oxidation reaction of ascorbic acid also occurs, but because the content of ascorbic acid in apples is very small, ascorbic acid oxidation is not the main factor causing browning. (2) Use transmission electron microscopy to observe changes in the ultrastructure of apple pulp cells with different degrees of browning. It was found that pulp browning is similar to changes in the ultrastructure of fruit senescent cells, with cell wall deformation, plasmolysis, and membrane structure disintegration. , chloroplasts transform into starch granules, mitochondrial inner cristae and membrane rupture, etc. As the degree of browning increases, plasmodesmata decrease in density, quantity, and orderliness, and in severe cases, breakage occurs. The ultrastructural disintegration of organelles such as vacuoles, plastids, and mitochondria and the plasma membrane of cells and the destruction of cell regional structures lag behind the browning of the pulp. Therefore, enzymatic browning caused by the destruction of cell regional structure is not the only way to cause browning of pulp tissue. Maillard reaction is a major form of early browning. (3) Research on the effects of different testing distances, measurement temperatures, measurement locations, surface colors, and storage times on the near-infrared spectral response characteristics of apples shows that the detection location, measurement distance, and storage time have a significant impact on the spectral response characteristics of apples. , collecting spectra at the equator of the fruit at zero distance can reduce errors; fruit temperature and surface color have no significant impact on the near-infrared spectrum, and the effects of temperature and surface color can be ignored in actual detection. (4) Comparatively analyzed the impact of the instrument's spectral collection parameters on the near-infrared spectral response characteristics of apples, and determined that the optimal parameters of the spectrometer are: the instrument signal energy is 5V, the scanning resolution is 8cm~(-1), and the number of scans is 64 Second-rate. Such optimized parameters can not only meet the actual testing requirements for apple quality, but also simplify operations, reduce costs, and improve testing efficiency. (5) The cluster analysis method based on Euclidean distance is used to judge and eliminate abnormal samples, and the most effective representative modeling samples are 170. The misjudgment rate of the built qualitative model is reduced from the original 53% to 29%; the analysis is different The influence of spectral preprocessing method on the qualitative prediction model of apple pulp browning. The qualitative model built after SNV combined with SD preprocessing has the best prediction effect, and the misjudgment rate is reduced to 22%; the effective spectral band for detecting apple pulp browning is obtained. Range: 10500cm~(-1)~6960cm~(-1), 6760cm~(-1)~5300cm~(-1), 5100cm~(-1)~4500cm~(-1), the stability built in this area The model misjudgment rate is reduced from 22% of the original full-band model to 19%; comparing the prediction accuracy of the qualitative model built by the standard algorithm and the factorization method, the factorization model has a high prediction accuracy and has fewer misjudgments on the modeling set and the test set. The rates are 16% and 19% respectively.

(6) Quantitative non-destructive testing of apple pulp browning degree is proposed. This method can not only detect the browning degree of apple pulp, but also achieve prediction and early warning of pulp browning. By effectively eliminating samples with abnormal physical and chemical detection and abnormal spectra, and then gradually eliminating similar samples, representative modeling samples were selected, and a quantitative detection model for apple pulp browning was established using PLS combined with MSC. The RMSECV of the model was 0.082, and the RMSEP The cross-validation coefficient of determination R~2 is 0.084, the cross-validation coefficient of determination R~2 is 0.871, and the external validation coefficient of determination R~2 is 0.853. It shows that the built model has high prediction accuracy and can make actual predictions for unknown samples. (7) It was found that the characteristic fingerprint wave numbers related to browning of apple pulp are 8822cm~(-1), 7085cm~(-1), 7000cm~(-1), 6694cm~(-1), 5800cm~(-1) , 5322cm~(-1), 4650cm~(-1), and used characteristic fingerprint wavenumbers to establish an MLR model for apple pulp browning detection. The RMSECV of the model is 0.077, the RMSEP is 0.079, and the cross-validation determination coefficient R~ 2 is 0.908, and the external verification coefficient of determination R~2 is 0.878. This method can simplify the detection operation, improve detection efficiency, and provide a fast, intuitive, simple and feasible detection method for apple pulp browning.