상세 보기
Prediction of Atmospheric Corrosion Current Considering Environmental Characteristics and Member Directionality
- Kwon, Tae-Yun;
- Park, Jun-Seo;
- Seong, Taek-Ryong;
- Huh, Jungwon;
- Ahn, Jin-Hee
WEB OF SCIENCE
0SCOPUS
0초록
Steel structures exposed to atmospheric environments undergo corrosion, leading to sectional loss and reduced durability. This study investigates a machine learning based framework for predicting the level of corrosion current occurring in steel structures, considering seasonal environmental variability and member directionality. Prediction models were composed using electrochemical corrosion current data from controlled indoor tests and one year of atmospheric exposure. Indoor tests examined the effects of temperature, relative humidity, chloride concentration, and member orientation, while atmospheric models reflected natural seasonal variability using daily maximum, minimum, and mean temperature and humidity, dew point, and temperature gradient variables. Random Forest and XGBoost models were trained and validated using an 80:20 data split. For indoor data, the models captured increased corrosion current with chloride concentration and consistent directionality effects, while cumulative corrosion current improved accuracy but limited general applicability. For atmospheric data, corrosion current was effectively predicted using environmental and directional variables, with feature importance varying seasonally. Validation with unseen data showed improved reproduction of long-term trends as training duration increased. Under low-pollution atmospheric conditions dominated by temperature and humidity effects, the proposed approach enables prediction of corrosion current trends using environmental data alone, without requiring direct corrosion current measurement, accounting for member directionality.
키워드
- 제목
- Prediction of Atmospheric Corrosion Current Considering Environmental Characteristics and Member Directionality
- 저자
- Kwon, Tae-Yun; Park, Jun-Seo; Seong, Taek-Ryong; Huh, Jungwon; Ahn, Jin-Hee
- 발행일
- 2026-09
- 유형
- Article; Early Access
- 언어
- ENG
- 출판사
- 한국강구조학회
- 발행국가
- 대한민국
- ISSN
- E 2093-6311
P 1598-2351