Evaluating the corrosion resistance of marine steels under different exposure environments via machine learning. (1st January 2023)

• Record Type: Journal Article
• Authors: Li, Zhuang; Long, Zhilin; Lei, Shan; Liu, Xiaowei; Yang, Lingming; Zhang, Wei; Zhang, Ting
• Is Part Of: Physica scripta. Volume 98:Number 1(2023)
• Publication Date: 2023-01-01
• Subjects: corrosion rate -- machine learning -- steel -- marine environment; Physics -- Periodicals; 530.05
• DOI: 10.1088/1402-4896/aca43a
• Languages: English
• ISSNs: 0031-8949

Abstract

Abstract: The corrosion behavior of marine engineering steels in marine environment is an extremely complex process, which poses great challenge to accurately evaluate the corrosion resistance of various stees in different marine environment. Owing to the wide application of machine learning (ML) approaches and the accumulation of corrosion data of different steels in natural marine environment, herein, we reported eXtreme Gradient Boosting (XGBoost) ML models for predicting the corrosion rate in submerged, tidal and splash zones. By taking material composition, environmental factors and exposure time as inputs, the developed prediction models can well predict the corrosion rate with the accuracy of 93%, 96% and 93% for submerged, tidal and splash zones, respectively. In addition, we identified the key factors affecting the corrosion resistance of steels in different marine zones, and analyzed the relationship between these factors and corrosion rate by applying SHapley Additive exPlanations (SHAP) method. This work demonstrates that ML model combined with SHAP method are efficient in evaluating corrosion behavior of various steels in different marine environment.