A critical analysis of machine learning in ship, offshore, and oil & gas corrosion research, part I: Corrosion detection and classification

Abstract

Corrosion poses a significant threat to the integrity and longevity of ship, offshore, and oil & gas structures, resulting in substantial economic losses, environmental hazards, and safety concerns. In recent years, machine learning (ML) has emerged as a promising tool for corrosion analysis in maritime industry. This paper provides a critical review of prevalent ML approaches, including Convolutional Neural Networks (CNNs), Random Forests (RFs), computer vision, image processing techniques, and hybrid models in corrosion detection and classification from 2018 to 2024. Beyond a typical review, this study meticulously examines these approaches, focusing on model development, efficacy, limitations, and practical implementation challenges in details. Key findings reveal that while ML models hold considerable potential to enhance the efficiency of corrosion detection and classification, significant barriers such as data quality, model interpretability, and integration into existing maintenance workflows impede widespread adoption. Furthermore, the paper identifies best practices and proposes future research directions to bolster the robustness and reliability of ML models in corrosion analysis. The insights gleaned from this review aim to guide industry experts and academicians in developing more effective corrosion management strategies through the integration of machine learning, ultimately mitigating the impact of corrosion on maritime and offshore operations.