Comparison of methods for wind turbine condition monitoring with SCADA data

Wind turbine operational costs can be reduced by monitoring the condition of major components in the drivetrain. SCADA-based condition monitoring is attractive because the data are already collected, resulting in rapid deployment and modest set-up cost. Three SCADA-based monitoring methods were reviewed: signal trending; self-organising maps and physical model. The physical model was identified as being the most reliable at predicting impending component failures. A validation study on this method using five operational wind farms showed that it is possible to achieve a high detection rate and good detection accuracy. An advance detection period of between 1 month and 2 years was achieved by the method. The study has also highlighted limitations and areas for further development.