Dynamic neural network modeling for hydrochloric acid recovery process

This paper describes neural network models for the prediction of the concentration profile of a hydrochloric acid recovery process consisting of double fixed-bed ion exchange columns. The process is used to remove the Fe²⁺ and Fe³⁺ ion from the pickling liquor, resulting in increasing the acid concentration for reusing in the pickling process. Due to the complexity and highly nonlinearity of the process, the modeling of the process based on the first principle is difficult and involve too many unknown parameters. Therefore, an attractive alternative technique, neural network modeling, has been applied to model this system because of its ability to model a complex nonlinear process, even when process understanding is limited. The process data sets are gathered from a real hydrochloric acid recovery pilot plant and used for neural network training and validation. Backpropagation and Lenvenberg-Marquardt techniques are used to train various neural network architectures, and the accuracy of the obtained models have been examined by using test data set. The optimal neural network architectures of this process can be determined by MSE minimization technique. The simulation results have shown that multilayer feedforward neural network models with two hidden layers provide sufficiently accurate prediction of the concentration profile of the process.