Drying kinetics of semi-finished cassava crackers: A comparative study

Drying kinetics of semi-finished cassava crackers was investigated in this paper using a hot air dryer at seven levels of drying air temperatures in the rage of 50–80 °C, a fixed air flow velocity using a fan speed of 0.18 kW, and a fixed level of thickness at 1.5 mm. A comparative study was performed among mechanistic and empirical models: the diffusion model, Newton model, Page model, Modified Page model, Henderson and Pabis model, MFNN (Multilayer Feedforward Neural Network), and ANFIS (Adaptive-Network-based Fuzzy Inference System), to estimate dynamic drying behaviors of semi-finished cassava crackers. Among these models, MFNN was found to be the most suitable for predicting moisture ratio of the product based on r2 (regression coefficient), and MSE (mean squared errors between the experimental data and predicted values).