A framework that alerts wind farm operators to failing components in offshore wind turbines and estimates their remaining useful life, based on signals from sensors.
Maintenance and operations account for 25–30% of an offshore turbine's total lifetime cost. Efficient planning of repair trips is crucial, as they depend on weather windows, and vessel and crew availability. Early warnings allow for better planning.
Two models work in parallel. The first model detects which component is failing. The second model estimates the remaining useful life of the failing component.
Tested on limited available data, the detection model raised alarms for all failures, but showed high false alarm rates. The remaining useful life model was far less accurate on held-out failures than on training data, highlighting insufficient generalization.