Wind turbines stream thousands of sensor readings every ten minutes, yet faults are still missed and engineers waste time chasing false alarms. Unplanned failures mean costly repairs, long downtime and lost clean energy, especially offshore, where getting a crew to a turbine depends on the weather. My PhD at Heriot-Watt University and the University of Edinburgh is about making AI-based fault detection something operators can actually rely on. I’m building a research testbed that learns each turbine’s normal behaviour from SCADA data and flags when it starts to drift.