Usage of Biomass Gasifier for Drying Soaked Paddy in a Reversible Airflow Flatbed Dryer: Artificial Neural Network Modelling

Drying of parboiled paddy requires a significant amount of energy due to its high moisture content and poor driers’ efficiency. To address this issue, biomass gasifier is used to dry the soaked paddy in a reversible airflow flatbed dryer (RAFD). In this study, artificial neural network (ANN) modelling was used to study the interactions between the drying parameters and drying performance as it can deal with non-linear and complex problems. Prediction of the head rice yield (HRY), drying time (DT), specific energy consumption (SEC), specific fuel consumption (SFC), and overall system efficiency (OSE) based on the outcomes of process parameters such as drying air temperature (DAT), drying air velocity (DAV), and bed height (BH) is vital in comprehending the gasification process. Feedforward artificial neural network (FANN) and nonlinear autoregressive exogenous (NARX) models were used to carry out the predictions with Levenberg–Marquardt (LM) training algorithm being most suitable for FANN with overall coefficient of determination, R² of 0.9990, and low MSE of 0.37, 1.14, 0.06, 0.04, and 0.76 in estimating HRY, DT, SEC, SFC, and OSE, respectively. In contrast, Bayesian regularisation (BR) for NARX model, with highest overall R of 0.7711 but it had high MAPE of 26.74%, 20.37%, and 21.84% in predicting DT, SFC, and OSE, respectively. Therefore, in this study, FANN model is better compared to NARX model in terms of R and errors.