Digital twin output functions for monitoring wind turbines

Project Description:

Digital twin technologies enables the prediction of real-world performance of wind turbines based on a digitally constructed counterpart. By increasing the precision of digital twin performance brought about from new research and developments in this area, the operational efficiency of physical turbines can be increased – which will assist with the growth of offshore wind contributions to the UK’s power supply.

This PhD project aims to develop new knowledge for the creation of digital twins that will not only benefit wind farms, but a range of other engineering structures. By expanding on the underlying mathematical framework for combining physics and data based models, and accounting for uncertainties present in physical twins, a better set of methods for developing quantitative techniques in the creation of digital twin output functions can be mapped onto quantities of interest in the physical twin. In utilizing these output functions, geometrical and behavioral complexity can be incorporated whilst working with a relatively low number of system outputs.

The novelty of this approach is found in the combination of (i) time evolving dynamics models that capture the best physics-based knowledge with, (ii) statistical models to represent uncertainty, and (iii) machine learning and optimisation techniques to support operational decisions and enhance the overall system knowledge. These elements in conflation will allow for greater accuracy in the creation of digital twins and measurement of outputs, and in structural health monitoring throughout the lifecycle of the physical asset from design, through operation, to end of life. Only by achieving this can the digital asset truly fulfil its role as a proxy for important design/operation and asset management decisions throughout the lifetime of the asset.