The aim of this EngD will be to leverage the potential of this rich data set by using data-driven methodologies and tools to develop prescriptive maintenance strategies for offshore wind farms, with an emphasis on industrial application. This aim can be broken down into three key objectives, each following on from the other.
Firstly, before any predictive maintenance strategies can be formed, failure predictions are required. The data will be used to investigate the possibilities of early fault detection in key components using SCADA and CMS systems, aiming to describe failure distributions and build predictive models to anticipate component failures.
The second aim will focus on using and developing existing O&M models in order to develop prescriptive maintenance strategies. These strategies will be based on predictions from data and will recommend different O&M strategies depending on the method of optimisation. These models will incorporate failure statistics and predictions developed in the first aim.
Finally, work done in the previous two areas will be combined to develop a digital representation of an offshore wind farm O&M strategy based on prescriptive maintenance. This tool will incorporate failure predictions and results from the maintenance model. It will highlight where failures are likely to occur and recommend maintenance strategy options depending on the chosen method of optimisation. This tool will act as a visual aid to describe the current and potential future O&M strategy of the farm. As new data is gathered from the turbines this can be fed into the model to continuously update it, creating a prescriptive O&M 'digital twin' of the wind farm.