Automatic Mooring-Line-Failure Detection Using Hull Motion Sensors and Machine Learning (ML) for 15MW Floating Offshore Wind Turbines (FOWTs)

Mooring line failure in Floating Offshore Wind Turbines (FOWTs) threatens safety, power production, and nearby structures. This study develops an automated detection framework that combines hull motion sensors with machine learning (ML). A synthetic dataset for the 15MW VolturnUS-S semi-submersible FOWT was created using OpenFAST under intact and failed mooring conditions across 79 wind and wave scenarios. The mean and standard deviation of six degree of freedom (6DOF) motions including surge, sway, heave, roll, pitch, and yaw were used as input features. Different sensor configurations such as inertial measurement units (IMU) and differential GPS (DGPS) were tested under added noise. Artificial neural networks (ANN), random forests (RF), and support vector machines (SVM) were evaluated with optimized hyperparameters. All models achieved excellent accuracy, with RF performing best. However, using raw IMU signals without displacement conversion reduces prediction accuracy due to missing mean offset information. The results demonstrate a reliable approach for near-real-time mooring-integrity monitoring that improves safety while reducing reliance on expensive equipment.