Physics-Informed Multimodal Sensing and Machine Learning Framework with Real-Time Digital Twin for Intelligent Structural Health Monitoring of Marine Wharves

Marine wharves in harsh sea conditions often develop early cracks and corrosion in their pile foundations. Traditional vibration‑based structural health monitoring (SHM) systems usually can’t tell whether changes are caused by the environment or by real damage. To overcome this, the research proposes a framework that combines several types of sensors—accelerometers, strain gauges, high‑resolution cameras, acoustic transducers, and corrosion sensors—with edge computing and a real‑time physics‑informed digital twin. Edge nodes clean and process the raw sensor data, extracting modal frequencies, acoustic features, and crack indicators. This preprocessing cuts the amount of data that must be sent over the network. The diagnostic pipeline then uses an attention‑based multimodal fusion mechanism, XGBoost

screening, and a feedforward deep neural network. The network is trained with a composite physics‑informed loss function that penalizes modal‑frequency shifts and stiffness ratios that fall outside the acceptable range of 0.70 to 1.00. By embedding physical constraints in the loss, the system stays reliable even when environmental or operational conditions vary. This method offers a solid foundation for both academic research and practical deployment of intelligent monitoring systems for marine infrastructure.

Message: Hi FE, I need help with “Physics-Informed Multimodal Sensing and Machine Learning Framework with Real-Time Digital Twin for Intelligent Structural Health Monitoring of Marine Wharves” in “Civil Engineering”