wcm.03.2025.455.460

Freshwater resources are being rapidly exhausted as a result of both natural and anthropogenic activities. In recent years, there has been considerable interest in the potential and opportunities presented by the desalination technology for brackish and seawater. especially reverse osmosis desalination to mitigate the increasing of water scarcity and supplying fresh water. In this research, the performance prediction of a multistage with two passes of medium-sized spiral wound brackish water RO (BWRO) desalination plant with capacity (1200m3/day) for Arab Potash Company (APC) placed in Jordan is developed and forecasted for one month by utilizing an artificial neural network (ANN) model as a smart manufacturing system. For predicting the next one-month values of permeate flowrate, recovery, and rejection of product water plant, a neural network such as Multilayer perceptron (MLP) and radial basis function (RBF) were developed and trained based on the feed water parameters which include pH, pressure, conductivity and Temperature to predict the values of permeate flowrate, plant recovery, and plant rejection for one-month. The results have been predicted and indicated that both forms of neural networks are extremely reasonable for forecasting permeate flowrate, plant recovery, and plant rejection. Forecasting of plant performance, with both Multilayer perceptron (MLP) and radial basis function (RBF) neural networks models, generated predictive results with good accuracy for long-term memory time intervals extended to 725 hr for permeate flowrate, recovery, and rejection of the product water plant for forecasting times up to one month. Up to this, this research would be an effective tool used for predicting the good desalination plant performance, which contribute to saving the cost and energy.