Intelligent process optimisation based on cutting-edge emotional learning for performance evaluation of NF/RO of seawater desalination plant

Abstract

As decision-makers, researchers encounter highly dynamic, complex problems requiring suitable nature-based and industrial quantitative tools for performance analyses, syntheses, and modelling. This study employed a new-generation emotional artificial neural network (EANN) and other stand-alone approaches to model the performance of hybrid nanofiltration (NF) and reverse osmosis (RO) desalination plants in terms of permeate recovery (PR) (%) and permeate flow rate (PFR) (m3/h). With the obtained experimental data, augmented Dickey-Fuller (ADF) and Philip Perron tests, evaluated using the Akaike information criterion and Schwarz information criterion, were employed to assess stability and stationarity. Nonlinear input-variable feature selection based on mutual information and neuro-sensitivity analysis (NSA) was conducted on several input variables before modelling. The EANN was compared with stand-alone predictive models: backpropagation neural network and classical stepwise regression. Predictive simulations were evaluated using the Nash–Sutcliffe model efficiency coefficient (NSE), correlation coefficient (CC), mean absolute error (MAE), determination coefficient (DC), root mean square error (RMSE), and two-dimensional Taylor graphical visualisation. For PR, in the calibration phase, the NSE and MAE (%) were 27–99 % and 0.0064–0.1410, respectively. For PFR, NSE and MAE (m3/h) were 50–99 % and 0.0064–0.1410, respectively. The bibliography review indicated the popularity of artificial intelligence (AI) applications in desalination. Results revealed that this new-generation EANN demonstrated superior accuracy over all models; accordingly, it can serve as a reliable tool for evaluating the performance of NF/RO hybrid seawater desalination plants.