Artificial Neural Networks in Corrosion Prediction and Modeling: A Critical Review of Advances, Limitations, and Future Prospects
Journal of Bio- and Tribo-Corrosion, cilt.12, sa.4, 2026 (Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 12 Sayı: 4
- Basım Tarihi: 2026
- Doi Numarası: 10.1007/s40735-026-01218-6
- Dergi Adı: Journal of Bio- and Tribo-Corrosion
- Derginin Tarandığı İndeksler: Scopus
- Anahtar Kelimeler: Artificial neural networks, Corrosion prediction, Data-driven modeling, Machine learning, Materials degradation, Microbiologically influenced corrosion, Tribo-corrosion
- Orta Doğu Teknik Üniversitesi Adresli: Evet
Özet
In many sectors, corrosion is a recurring and expensive problem where precise prediction is challenging due to intricate relationships between material qualities, environmental factors, and time-dependent degradation. Conventional corrosion prediction techniques, which rely on statistical regression, empirical formulas, and electrochemical modeling, frequently fail to adequately capture the intricate, nonlinear relationships between environmental factors, material characteristics, and electrochemical parameters. Thus, Artificial Neural Networks (ANNs), a branch of artificial intelligence that draws inspiration from the learning process of the human brain, have become a viable computational method for accurately and adaptively forecasting corrosion behavior. In the review, various ANN architectures, including feedforward, backpropagation, gradient descent, radial basis function, and deep learning networks, are highlighted, along with their performance in modeling intricate corrosion systems. This paper also looked at the use of ANNs for corrosion prediction in a wide range of environments, such as reinforced concrete, pipelines, coated materials, and air exposure. The review further examines ANN applications in microbiologically influenced corrosion and in tribo-corrosion. It evaluates physics-guided neural networks, symbolic regression, transformer architectures, explainable AI methods and uncertainty quantification, which address the interpretability and data scarcity limitations of conventional ANN models. This review identifies Artificial Neural Networks as a transformative tool in corrosion research, providing predictive accuracy and flexibility that traditional approaches cannot match. But in order to reach their full potential, future studies should focus on explainable AI frameworks, big, high-quality datasets, and ANN integration with mechanistic corrosion models. Data-driven asset management and proactive corrosion control in industrial applications will be made possible by such interdisciplinary approaches, which will enhance the scalability, transparency, and dependability of corrosion prediction systems. Graphic Abstract: (Figure presented.)