Variation Law and Predictive Modeling Construction of Internal Quality in Korla Fragrant Pears Under Multi-Type Damage During Storage

To enhance the economic benefits of the Korla fragrant pear industry and reduce fruit loss rates, this study investigates changes in pear fruit quality during storage under different damage types. An adaptive neuro-fuzzy inference system (ANFIS), tested with eight different membership functions, was used to predict the hardness and soluble solids content (SSC) of pears stored under various damage conditions. The results showed that both hardness and SSC of damaged pears decreased with prolonged storage time. During identical storage periods, more severe damage led to faster declines in hardness and SSC, among which impact loads and combined impact-compression loads caused the most rapid reductions in pear hardness and SSC. The ANFIS model with the gauss2mf membership function achieved optimal prediction accuracy for pear hardness (RMSE = 0.2207, R2 = 0.9434); the ANFIS model with the trimf membership function yielded the best prediction for pear SSC (RMSE = 0.2016, R2 = 0.9701). This study provides a theoretical basis for optimizing warehouse management and quality control of Korla fragrant pears, and for determining the optimal treatment window during storage.