Modeling physicochemical characteristics of Apple using adaptive neuro-fuzzy inference system

Abstract

Apple is considered a major crop in the Iranian agriculture market. Cold storage allows for the preservation of this fruit for an extended period. However, the qualitative characteristics of apples can be affected by their internal changes during storage. Therefore, we must predict these changes and provide suitable storage conditions to maintain the nutritional and economic values of this crop. This study analyzed the physicochemical characteristics of Golden Delicious apples under storage at 0 °C and 4 °C for 0, 45, 90, and 135 days. The examined physicochemical characteristics were pH, firmness, density, soluble solids (SS), and moisture. The adaptive neuro-fuzzy inference system (ANFIS) was then employed to predict the physicochemical characteristics of apples based on color space components (L*a*b*), CT (Computed Tomography) number and storage temperature and duration. The results of implementing and comparing different ANFIS models indicated that R², RMSE, MAPE, and EF in the best models of prediction were 0.954, 1.793, 3.580, and 0.910 for firmness, 0.965, 0.085, 1.565, and 0.931 for PH, 0.970, 0.026, 2.422, and 0.940 for density, 0.960, 0.309, 1.349, and 0.921 for SS, and 0.980, 0.0005, 0.448, and 0.960 for moisture, respectively. As per the results, we can accurately and roughly predict the physicochemical characteristics of apples under cold storage to assess quality during storage.