Intelligent Structural Health Monitoring of Jack-Up Platform Legs Using High-Density Sensor Networks, Real-Time Digital Twins, and Machine Learning-Based Damage Detection

Abstract

Offshore jack-up platforms operate in harsh marine environments, where undetected structural degradation can quickly escalate into catastrophic failure. To address this challenge, this study presents an intelligent Structural Health Monitoring (SHM) framework specifically designed for jack-up platform legs. Unlike previous offshore SHM approaches that often rely on single-modality vibration analysis or lack field-validated digital twins, the proposed system integrates high-density wireless sensor networks, a real-time physics-based digital twin, and advanced machine learning to enable accurate and timely damage detection under real operating conditions. The architecture deploys 128 marinized sensor nodes in a mesh network, continuously feeding data into a high-fidelity finite element digital twin updated via extended Kalman filtering. A hybrid learning pipeline—combining variational autoencoders for anomaly detection with graph neural networks for spatial localization—enables high-resolution damage assessment without extensive labeled field data. Laboratory-scale validation on a 1:22 physical model and preliminary offshore deployment demonstrated the system’s ability to detect stiffness losses as small as 2.5%, with localization errors within ±0.8 m and a classification accuracy of 94.2%. Edge computing at the sensor level reduced communication loads by 65% while maintaining an end-to-end inference latency of 850–1100 ms. Field deployment over six months achieved 94.7% uptime, with maintenance costs reduced by 23% and unplanned downtime by 41%. An extended 18-month trial further achieved 99.2% availability, a 34% cost reduction, and a 67% reduction in downtime. This integrated, field-proven framework offers a scalable solution for continuous monitoring, early damage detection, and maintenance optimization in critical offshore assets.