Artificial neural network for predication of zinc consumption rate of cathodic protection of copper in saline water: A short communication

Abstract

The rate of zinc consumption (anode) for cathodic protection of copper pipeline in saline water is forecasted using artificial neural network (ANN). Zinc consumption rate (input dependent variable) is estimated as a function of temperature, flow rate, time and NaCl concentration (output independent variables). One hundred ANNs are created using STATISTIC 10 software based on the Intelligent Problem Solver tool. Only ten high performance networks are retained. Three types of ANNs are constructed. Linear Model (LM), Multi-Layer Perceptrons (MLP), and Radial Basis Function (RBF). MLP 4:4-7-1:1 with four input variables and one output variable, and three layers of 4, 7, and 1 unit, respectively is the high performance network with a 0.9946 correlation coefficient and 0.0071 absolute errors. The predicted results are in a good agreement with experimental one. It is found that the rate of zinc consumption increases with increasing of all independent variables. Sensitivity analysis showed that the time is the most sensitive variables, while salt concentration, flow rate, and the temperature had the lower effect, respectively.