Nonlinear Modeling of Chromium Tanning Solution Using Artificial Neural Networks
In this article artificial neural networks (ANNs) are applied for multivariate calibration using spectroscopic data and for generation of quantitative estimates of the concentrations of a component (chromium) in solutions. Neural networks are capable of handling nonlinear relationships. Absorbance is nonlinearly dependent on concentration, especially in the case of wide concentration ranges and multicomponent solutions. In addition to the aforementioned reasons, nonlinearities are also caused by aging and by differences in pH and in the temperatures of the chromium-tanning solutions to be modeled. The sigmoid output function was used in the hidden layer to perform nonlinear fitting. The results are compared with the results obtained with principal component regression (PCR) and partial least-squares regression (PLS) methods.