Application of artificial intelligence hybrid models in safety assessment of submarine pipelines: Principles and methods

Submarine pipeline is a critical component for exploiting marine oil and gas resources. The aging of pipelines is becoming increasingly serious, and any damage or rupture can lead to severe marine environmental pollution. Therefore, effective safety assessment methods are crucial for ensuring the safety of submarine pipelines. Traditional safety assessment methods such as Event Tree Analysis (ETA), Fault Tree Analysis (FTA), and Failure Mode and Effects Analysis (FMEA) have been widely used but rely heavily on historical accidents. Artificial intelligence methods, particularly hybrid models, offer a more resilient alternative by combining different techniques, providing more accurate assessments even with limited data. Therefore, this study reviews the application of AI methods in the safety assessment of submarine pipelines, focusing on hybrid models that combine Support Vector Machine (SVM), Bayesian Network (BN), and Artificial Neural Network (ANN) with other techniques. These hybrid models have shown excellent performance in detecting and predicting corrosion defects in submarine pipelines and overall safety assessment by complementing each other's strengths. Future research should focus on model integration and optimization, revealing the evolution processes of accidents, data augmentation to address data scarcity, interdisciplinary research to enhance model interpretability, and intelligent decision-making for better pipeline integrity management.