Application of adaptive neuro-fuzzy inference system (ANFIS) in predicting quality characteristics of stored apple fruit

Quality evaluations of apple fruit during storage can help producers to choose and optimize suitable storage conditions. The internal changes of this product during the storage period will be caused change in the quality characteristics of the fruit. Prediction of these changes and creating suitable storage conditions are important steps towards maintaining the nutritional and economic value of the product. In this research, some physicochemical characteristics of Golden Delicious apples were measured during storage at two temperatures of 0 and 4 °C for 0, 45, 90 and 135 days. These characteristics included CT number obtained by X-ray imaging, pH, firmness, density, total soluble solids index and fruits moisture content. Then, using adaptive neuro-fuzzy inference system (ANFIS) the mentioned characteristics changes during storage were determined and compared with the actual values. The ANFIS model inputs were color components L*, a* and b*, storage temperature and storage duration, and the outputs were the mentioned physicochemical characteristics. According to the results, in the best selected models, the values of R2, RMSE, MAPE and EF statistical parameters for CT number were 0.909, 24.331, 11.319% and 0.899, for fruit firmness were 0. 950, 1.862 N, 3.298% and 0.904, for pH value were 0.912, 0.134, 1.134% and 0.839, for density were 0.910, 0.045 g/cm3, 7.223% and 0.828, for soluble solids were 0.884, 0.537% Brix, 2.340% and 0.781 and for fruit moisture content were 0.945, 0.008, 0.729% and 0.893, respectively. These results show that it is possible to predict the characteristics of apple fruit under storage conditions with high accuracy. This prediction will be useful to determine and maintain the quality of the product during storage.