Modelling some quality attributes of a convective Hot-Air dried tomato slices using ANN and ANFIS techniques

Abstract

The study investigated how different processing combinations affect the quality of tomatoes dried in a convective hot-air dryer. The Taguchi technique was used to plan the experiments. Three pretreatment methods were used: water blanching (WBP), ascorbic acid (AAP), and sodium metabisulphite (SMP). The slice thickness was changed from 4 to 6 mm, and the air temperature was changed from 40 to 60 °C. Standardised protocols were followed to assess the quality attributes, percentage shrinkage, rehydration ratio, as well as the levels of lycopene, β-carotene, and ascorbic acid in the dried tomatoes. The artificial neural networks (ANN) and adaptive neuro-fuzzy inference systems (ANFIS) models were trained using the data. At the best conditions of SMP, 6 mm slice thickness and 40ᵒC air temperature, the quality attributes were; 90.89 %, 4.22, 10.74 mg/100 g, 9.14 mg/100 g, and 25.14 mg/100 g, respectively. The findings demonstrate that ANN and ANFIS models provide a more accurate prediction. The ANFIS model, on the other hand, has proven to be more effective, with a greater coefficient of determination (≥ 0.9988) and lower root mean square error (≤ 0.02076) and mean absolute error (≤ 0.01623). The predictive models were experimentally verified to be accurate when compared to experimental results.