ABSTRACT
In recent years, numerical analysis of corrosion characteristics to accurately predict corrosion progress has received increasing attention from researchers. In actual construction, steel is mostly treated with coating for corrosion prevention. This study used steel SS400 with two kinds of coating, they were taken from the actual steel bridges. Defects were created artificially on the coating to simulate the corrosion progress when the coating is defective. Corrosion tests were carried out in two corrosive environments. Corrosion data of steel plate surfaces at different stages were used as data sets. In this study, a generative adversarial network (GAN) was used to build a prediction model of the corroded surface. The prediction model could simulate the damage of paint-coating around the defect at various stages and predict the corrosion of the steel plate in the final removal of the coating. Based on comparative results, this corrosion prediction model showed outstanding performance.
