73 Modeling of Ion Exchange Process Using Time Delayed Neural Networks

This paper describes a neural network model for the up-take of copper during an ion-exchange process. Since the ion-exchange process is a complex and nonlinear process, the modeling based on time related empirical equations is simplistic and difficult to account for all the concomitant processes influencing the process variables. Thus, the time delayed neural networks is used to model this process because of its ability to model complex nonlinear systems without fully understanding the system. The minimum square error (MSE) minimization technique is used to determine the optimal neural network architecture for the process. The simulation results show that time delayed neural networks with two delayed inputs and one hidden layer is sufficient to predict concentration of copper during ion exchange.