Apparent viscosity prediction of carboxymethyl-cellulose polymer aqueous solutions using artificial neural networks and regression techniques

The statistical analysis and artificial neural networks were used to estimate apparent viscosity of carboxymethyl-cellulose aqueous solutions. Apparent viscosity was measured by using a coaxial cylinder viscometer with six diverse shear rates over a wide range of temperature (27-54°C) and concentration (0.66-1.97 wt. %).First, an artificial neural network (ANN) was developed for the prediction of the apparent viscosity. The developed model was based on a three-layer neural network with 3 neurons in the hidden layer and a feedforward learning algorithm. The neural network was trained with binary systems consisting of 504 data sets and using shear rate, concentration combined with temperature as the input. A comparison of the experimental values and the results predicted from the neural network revealed a satisfactory correlation, with the overall absolute average deviation (AAD) of 0.13 and least square of 0.997 for apparent viscosity. The results were further compared to a generalized model as an alternative empirical model where it gained high AAD and low least square value. The neural network produced better results than the generalized model

Keywords:

Artificial Neural Networks , Carboxymethyl Cellulose , Apparent Viscosity , Empirical Mod