Machine Learning strategieën voor het verbeteren van computationele efficientie voor windmolenpark modellering

Current operational wind resources forecasting models (e.g. used for market bidding strategies or planning of ancillary services), do not consider effects of two-way coupling in large off-shore wind farms. With modern wind turbines reaching capacities of up to 9.5 MW and heights of up to 220m, and considering the planning of large wind farm clusters in Europe's North-Sea region, the inclusion of these effects is important. Currently, neither planning tools, nor operational forecasting tools take two-way meso-scale coupling into account. As a case study, the Belgian offshore zone and Belgian's offshore wind farms are considered. This PhD will work on developing the algorithm for the forecasting of the wind resources. The algorithm will be based on the operational weather forecasts adjusted for the effect of feedback processes using mesoscale atmospheric model integrations at a resolution of 0.3 km together with an atmospheric pertubation model (developed by the KU Leuven). Ultimately, this PhD aims to develop a tool for wind resource forecasts ranging from 1 to 7 days.