Development of a Physics-Informed Neural Network Model for Real-Time Structural Health Monitoring of Offshore Oil and Gas Platforms in the Niger Delta

Structural health monitoring of offshore oil and gas platforms in the Niger Delta operates in a data-rich but model-sparse environment where classical physics-based damage detection approaches are computationally prohibitive for real-time application and pure data-driven methods lack the physical interpretability and reliability required for safety-critical offshore structural decisions. This research develops a novel physics-informed neural network modelling methodology that embeds governing structural mechanics equations as soft constraints within deep learning architectures to enable rapid, interpretable, and physics-consistent damage state estimation in offshore jacket platform structures. The architecture incorporates the finite element equations of structural dynamics as physics residual loss terms alongside training data from both numerical simulations and scaled experimental models representative of Niger Delta platform configurations. Transfer learning strategies are developed to enable effective adaptation of models trained on well-characterised platform configurations to operational platforms with incomplete structural documentation. The network is trained on a dataset of 12,000 simulated damage scenarios spanning single and multiple simultaneous member degradation conditions, encompassing corrosion thinning, fatigue crack growth, and weld joint deterioration. Experimental validation uses a scaled physical model subjected to wave loading in a controlled laboratory facility, with damage states introduced progressively. The physics-informed network demonstrates detection accuracy of 94 percent for damage extents exceeding 15 percent section loss and localisation accuracy within one structural bay, outperforming purely data-driven baselines by 12 percentage points while requiring 74 percent less training data. Keywords: physics-informed neural network, structural health monitoring, offshore platform, Niger Delta, damage detection.