Towards a Population-based SHM: A Case Study on an Offshore Wind Farm

The use of offshore wind farms has been growing in recent years, mainly because of the advantages they present when compared to their onshore equivalents. However, the cost of an offshore wind farm is relatively high, and therefore their reliability is crucial if they ever need to be fully integrated into the energy arena. A wind farm consists of multiple structures which are meant to be nominally identical, so they form an ideal candidate for an application of population-based SHM. This paper presents the use of supervisor control and data acquisition (SCADA) extracts from the Lillgrund wind farm for the purposes of structural health monitoring. A machine learning approach, based on Gaussian Processes (GPs), is applied in order to model individual wind turbines, and then predict measurements of each wind turbine from the measurements of other wind turbines in the farm. Subsequently, appropriate and robust thresholds are applied in control charts in order to monitor the whole farm in a population-based approach, meaning that the predicted model from each turbine is used to monitor all the rest of the turbines. It is shown that novelty detection is possible with a low rate of false positives, something which paves the way for the practical application of SHM in a population of nominally identical structures.