Corrosion Risk Assessment in Coastal Environments Using Machine Learning-Based Predictive Models.

Area of Science:

• Environmental Science
• Materials Science
• Civil Engineering

Background:

• Atmospheric corrosion, driven by chloride deposition from marine aerosols, significantly impacts coastal infrastructure durability.
• Existing corrosion assessment standards rely on long-term data, hindering early-stage design evaluations.
• Increased coastal populations and industrial activity necessitate advanced corrosion prediction methods.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting chloride deposition levels.
• To identify key climatic and geographical parameters influencing atmospheric chloride deposition.
• To provide a tool for early corrosion risk assessment in coastal infrastructure.

Main Methods:

• Utilized machine learning algorithms: gradient boosting, support vector machine, and neural networks.
• Trained models on a dataset incorporating land coverage, wind speed, and orientation.
• Evaluated model performance using metrics such as the F1 score.

Main Results:

• Gradient boosting, a tree-based algorithm, achieved the highest prediction accuracy (F1 score: 0.8673).
• Identified influential environmental variables affecting chloride deposition.
• Demonstrated the effectiveness of machine learning in estimating chloride deposition.

Conclusions:

• Machine learning models offer accurate and efficient prediction of chloride deposition for coastal infrastructure.
• The developed approach supports corrosion monitoring and structural life assessment.
• This method provides a scalable and cost-effective solution for managing corrosion risks.