The application of the experimental design and the artificial neural network to the UV/H2O2 decolouration process of textile metal-complex dyes

Abstract

The UV/H2O2 decolouration of model dye solutions of three metal-complex dyes is described. For the first time, Plackett-Burman and full experimental design as well as artificial neural network (ANN) are applied to optimize the decolouration process of this class of textile dyes. After selecting parameters having an influence on the decolouration process, the corresponding range of intervals is determined. Results according to the experimental design show that the most efficient experiments are the combination of maximum and nominal parameter levels. With lower power and shorter time, very high decolouration efficiency can be obtained, which allows for minimising the costs of decolouration process. ANN data processing indicates very high cross-validation values, and the predictions of the values not included in the experimental design are in accordance with our expectations.