Modeling drying kinetics of tomato slices under convective hot-air using Artificial Neural Network (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) models

Modeling drying kinetics of tomato slices

Abstract

This study aimed to model the drying kinetics (Drying Time (DT), effective moisture diffusivity (D_eff), and Specific Energy Consumption (SEC)) of tomato slices dried with a convective hot-air dryer using Artificial Neural Networks (ANN) and Neuro-Fuzzy Inference System (ANFIS). The tomatoes were pretreated with Water Blanching (WBP), Ascorbic Acid (AAP) and Sodium Metabisulphite (SBP); sliced into 4, 6 and 8 mm thickness and dried at 40, 50 and 60^oC air temperatures. The experimental drying data were fitted to ANN and ANFIS models, while the best topology was obtained. The model's predictive performance was determined using the coefficient of determination (R²), Means Squared Error (MSE), Root Means Squared Error (RMSE) and Mean Absolute Error (MAE) between predicted and experimental results. The DT ranged between 11.5 and 22.5 h, D_eff (0.98 x 10^-10 to 6.36 x 10^-10 m²/s) and SEC (0.6247 to 1.9514 kWh/kg). Higher R² (0.9056–0.9834) with lower MSE (0.0014–2.2044), RMSE (0.00035 – 1.49 x 10^-13) and MAE (0.00026 – 1.08 x 10^-13) for ANFIS compared to ANN showed that ANFIS methodology could precisely predict experimental data. This study found that ANFIS is highly accurate in predicting the drying kinetic, thereby demonstrating its ability in finding a meaningful relationship between drying kinetic and drying conditions. Therefore, the model developed in this study can be a valuable tool in accurately predicting drying kinetic in dried tomatoes.