Ship hulls operate under repeated wave impact, changing loading conditions, and prolonged exposure to seawater. Over years of service, these effects can produce corrosion, local pitting, and fatigue cracks in important structural members. In practice, hull condition is still assessed largely through scheduled drydock surveys and empirical wastage relationships. Such inspections are valuable, but they provide only limited information about what happens between two survey periods. This paper develops a hybrid deep-learning approach that combines a Physics-Informed Neural Network (PINN) with a SpatioTemporal Graph Neural Network (ST-GNN) to estimate the future development of hull defects. Instead of relying only on measured inspection data, the model also uses corrosion behavior and Linear Elastic Fracture Mechanics as physical constraints during training. Measurements from different nondestructive testing sources, including ultrasonic gauging and autonomous underwater vehicle observations, are brought into a common structural representation. The evaluation described in this work uses 15 years of inspection data from 142 compartments of a Post-Panamax bulk carrier. Over a five-year prediction period, the reported corrosion-wastage MAE is 0.089 mm, while the RMSE for fatigue-crack propagation is 0.94 mm. The reported results show a 71.4% improvement over the recurrent baseline considered in the study. The physical constraints also prevent unrealistic outcomes such as an increase in plate thickness without repair or a negative crack-growth rate. The resulting workflow can be used as a computational aid for condition-based maintenance and continuous hull assessment.

Mrs.B.Moohambigai, S.Krishna Kumar, I.Sayed Arman, T.Thangaraj, S.Azizbasha,"AI-Driven Hull Defect Evolution Prediction: A Hybrid Physics-Informed Deep Learning Framework for Maritime Structural Health Monitoring" International Journal of Advanced Multidisciplinary Research and Educational Development, V2(5): Page(601-605) September - October 2026. ISSN: 3107-6513. www.ijamred.com. Published by Scientific and Academic Research Publishing.