Data-Driven Learning for Wind Energy Applications
For the wind energy, the offshore and motion control community, model-based design becomes more and more important due to ever increasing performance specifications and complexity. Model-based synthesis necessitates a nominal description of the real system, which can be derived from physical principles or measurement data. In the data-driven control cycle (see Fig.1 for an illustration),
the first step is to identify a model from measurement data and combine that with your first-principles model. Then the next step is to use this model to (re)design a controller. Followed by the implementation on an actual system and if necessary iterate or continuous execution of the data-driven cycle (learning).
Research statement: The development of efficient data-driven algorithms for the synthesis and analysis of robust controllers/estimators for large-scale mechatronic systems such as wind turbines and wind farms and evaluate them in a realistic environment (e.g. field testing, wind tunnel experiments, and high fidelity simulations).
With our ongoing research projects in the field of wind farm control, wind turbine control, wind turbine installation techniques, and floating wind we will contribute to the development of smart wind farms which will maximize the value of wind energy
Ongoing research projects
- Floatech (H2020)
- HKN: Wind Farm Flow Control (industry)
- Floating wind turbine installation (RVO)
- Watereye (H2020)
- Several Industrial projects
- VIDI (NWO): Active wake steering within densely spaced wind farms
Wind turbines in packed offshore ‘farms’ hinder each other, which lowers their efficiency. Researchers will develop robust control algorithms which actively steers each turbine’s wake away from other turbines. A novel integrated design will simultaneously optimize the operation strategy and farm topology, considerably reducing the cost of wind energy. - Farmconners (H2020)
- IEA Task 44: Wind Farm Flow Control [Operating Agent] (IEA)
- CRADA (NREL)