The demand for sufficient data in high quality to support the development of data-driven condition monitoring (CM) methods for offshore wind turbines has been dramatically increasing. However, the most commonly used data source in offshore wind, supervisory control and data acquisition (SCADA), technically contains limited information representing the characteristics of power electronic converters, from the aspects of operating points, aging processes, and failure modes.
In this project, a synthetic database will be generated jointly from physics-based simulation models and machine learning methods. The data quality and its practicality in supporting the development of CM for offshore wind turbine converters will also be examined.
