The energy sector has undergone a noticeable transition toward renewable and green energies as a result of the need to reduce carbon emissions in order to reduce human impact on climate change. One of the most promising and environmentally friendly energy-producing technologies is offshore wind.

Offshore wind turbines are subject to extreme mechanical stresses and are vulnerable to erosion because they operate in harsh marine environments. Although necessary for the wind turbine to function, the wind itself can have a negative effect, causing it to shut down when the wind speed exceeds 25 m/s. These reasons make dealing with technical issues a time-consuming and expensive task. In this situation, Omexom Offshore, a VINCI company with expertise in project development as well as the operation and maintenance of offshore facilities, made the decision to create a new AI solution in collaboration with the Leonard AI program to detect generator failures in advance. For this, the SCADA system’s already-collected sensor data were utilized.

To simulate each wind turbine’s typical behavior, periods during which it operated normally were used. This model enables the reconstruction of the signals’ typical behavior over time. The actual signal’s deviation from the reconstruction of normal conditions is then used to identify abnormal circumstances and, when the difference is large enough, to sound an alarm. Following a “Human-in-the-Loop” methodology, the triggered alarm is then sent to the operational manager for validation through a user-friendly interface. The effectiveness of the findings convinced Omexom Offshore to replace a generator in advance for the first time.