Artificial neural network modelling to predict the efficiency of aluminium sacrificial anode

Abstract

ABSTRACT Study explores the potential of a deep learning-based approach for predicting the current efficiency of aluminium sacrificial anodes in marine environments. The model takes into account various input variables, including the chemical composition of the sacrificial anode, pH, dissolved oxygen (DO), temperature, pressure, cathode electrode, current density, and the ratio of the surface area of the cathode to anode, with the anode current efficiency serving as the output variable. Utilising artificial neural networks in this study shows a mean absolute percentage error of 6.4% and 7.8% for the training and validation for predicting the current efficiency. The proposed model shows promising potential to predict the current efficiency of aluminium sacrificial anodes and improve the design of cathodic protection systems based on aluminium sacrificial anodes.