Abstract
Solar still productivity (SSP) essentially describes the performance of solar still systems and is an important factor to consider in achieving a reliable design. This study presents the use of artificial neural networks (ANN), adaptive neuro-fuzzy inference systems (ANTIS), and multiple regression (MR) for forecasting the SSP of an inclined solar still in a hot, arid environment. The experimental data used for the modelling process included meteorological and operational variables. Input variables were relative humidity, solar radiation, feed flow rate, and total dissolved solids of feed and brine. The models were assessed statistically using the correlation coefficient (CC), root mean square error (RMSE), overall index of model performance (01), mean absolute error (MAE), and mean absolute relative error (MARE). Overall, ANN was shown to be superior (CC = 0.98, RMSE = 0.05 L.m(-2).h(-1), OI = 0.95, MAE = 0.03 L.m(-2).h(-1), and MARE = 8.92%) to ANTIS and MR for SSP modelling. The relatively low errors obtained by the ANN technique led to high model predictability and feasibility of modelling the SSP. Thus, our findings indicate that ANN can be applied as an accurate method for predicting SSP.