Artificial neural network to predict the performance of the phosphoric acid production

The performance of the phosphoric acid production process is based on an increasing amount of heterogeneous data. This performance depends on several parameters that must be controlled at the unit of attack and filtration of the process. From the description of the attack unit, we found that a considerable flow of data can be generated that needs to be integrated for a parametric sensitivity analysis. After processing these data, we followed the artificial neural network method. This method allowed us to develop a model that predicts the yield of phosphoric acid at different operating parameters, with a very satisfactory error. Then the reliable model was exploited to evaluate the relative importance input parameters on the phosphoric acid production.