Modeling of brick obtaining proces with artificial neural networks

Abstract

In this study, neural network models were developed to predict the amount of NO discharged into the furnace chimney of a brick factory. The best performances were obtained with the MLP (6:18:6:1) model. Thus, in the training stage, the correlation coefficient was 0.9626 and the standard deviation was ± 5.61 mg/m3 and in the validation phase, a standard deviation of ± 15.24 mg/m3 is obtained. The advantages of this study derive from the important savings of time, materials and energy obtained by reducing the number of test loads in the analyzed industrial process.