第26回応用力学シンポジウム

Presentation information

Regular Session

General Session (1.Mathematical analysis for mechanics problem: forward- and inverse-modeling in civil engineering)

1.Mathematical analysis for mechanics problem: forward- and inverse-modeling in civil engineering

Sun. May 28, 2023 8:30 AM - 9:45 AM A (6号館 4階 6410室)

座長:珠玖 隆行(岡山大学)

9:15 AM - 9:30 AM

[21001-05-04] A GAN-based Prediction for Corrosion Progress of Paint-coated Steel with Different Defects (Proceedings of Symposium on Applied Mechanics)

*Feng Jiang1, Ayato Hamada1, Mikihito Hirohata1 (1. Osaka University)

Keywords:Paint-coated steel, Corrosion, Deep learning, GAN

A series of experiments and numerical analyses were conducted for proposing a prediction method for the corrosion progress of paint-coated steel. Two types and different sizes of defects were artificially created on the paint-coating on SS400 steel to simulate the corrosion after the coating was damaged. The corrosion tests were carried out under the accelerated corrosion test ISO 16539 Method B. A generative adversarial network (GAN) based prediction model was used to simulate the corrosion progression at the coating defects. The proposed model could predict the corrosion at the final stage around the coating defects on the steel plate surface.