Evaluating the fluidized-bed drying of rice using response surface methodology and artificial neural network

Abstract

In this work, the effect of fluidized bed dryer under different fluidization regimes and temperature on drying time and quality properties of rice has been investigated. To predict the outputs, a feed-forward multilayered perceptron and a central composite design were created. The chosen final artificial neural network (2-8-7-5) was compared to the response surface methodology for its modeling and predictive skills. Capturing the system's nonlinear behavior and simultaneous performance prediction demonstrate the superiority of a response surface methodology. The experiments proved that fluidized bed dryer with ventilation significantly increased the head rice yield (from 30 to 505%) and the whiteness index (from 2 to 17%) compared to fluidized bed dryer without ventilation. After that, we optimized the drying conditions by response surface methodology. The best optimum conditions are related to lowest drying time and the highest quality parameters. The optimal drying conditions are as follows: 52 °C inlet temperature, 3.1 m/s inlet fluidization velocity (bubble fluidization regime), and fluidized bed dryer with ventilation. At this optimum condition, the values of experimental test were found to be 218 min (drying time), 74% (head ice yield), 60.7 (whiteness index), 3.15 (water uptake ratio) and 1.84 (elongation ratio) with desirability factor of 0.695.