Mansoureh Yeganeh - Facuity Of Engineering, Department Of Chemical Engineering, Quchan University Of Technology,Iran
Soheil Hamidi Tabrizi - Facuity Of Engineering, Department Of Chemical Engineering, Quchan University Of Technology,Iran
Bahareh Tanhaei - Facuity Of Engineering, Department Of Chemical Engineering, Quchan University Of Technology,Iran

Artificial neural networks are a powerful and modern modeling tool, especially in cases where the relationship between data is unclear.Inspired by the way the biological nervous system works to process data and predict output responses in complex situations, and considering that the removal of color pollutants from industrial effluents is a serious environmental issue in the world, modeling these processes for Predicting the behavior of changing effective parameters can be useful. In this study, using in vitro results of tartrazine dye removal by adsorption and adsorbent of montmorillonite (MT) modified by surfactant hexadecylamine (HDA) and the effect of surfactant concentration and adsorption concentration, pH, temperature and Time, modeling and optimization with artificial neural networks. In the neural network designed using the error propagation algorithm and the reduction gradient method as a learning method and with the Adam optimizer function, the multilayer perceptron model has a data-based approach, the best Relu activator function with three hidden layers with ۴۰,۴۰ and ۳۰ neurons, with ۱۲۰۰ epoche were determined and the best results were obtained with correlation coefficient of ۰.۹۴ and mean squares error of ۰.۰۰۰۴. The results showed an acceptable correlation between neural network modeling and laboratory results.

Artificial Neural Networks, Neurons, Multilayer Perceptron, Montmorillonite