An artificial neural network approach for corrosion assessment of structures exposed to marine atmosphere

Abstract

The aim of this paper is to demonstrate the use of Artificial Neural Network (ANN) to assess the relative corrosion state of steel rebar in three different cementation materials (i.e. Ordinary Portland cement (OPC), Portland Pozzolona Cement (PPC) and Portland Slag Cement (PSC)). Steel rebar is embedded in M30 and M40 mix design with the cover thickness of 25 and 40 mm and exposed in corrosive marine environment for the accurate representation of real conditions. Open Circuit Potential (OCP) measurements were used for the analysis and model development through ANN. Comparison of measured and predicted OCP is done. It is very clear that the results obtained using ANN has high accuracy. Further, assessing the relative performance using ANN assures confidence in future predictability and is practical and beneficial.