The economical feasibility of the wind energy industry greatly depends on the reliability and maintainability of new large scale wind parks/wind farms. Especially for offshore wind farms where the distributed turbines are connected into a network, the topic of predictive maintenance is crucial in the explosive growth. On a wind farm, individual optimization of each turbine is now mainly done temporally, taking the local (temporal) changes of the wind into account. The optimization on a global scale is of interest and has been investigated in a centralized framework. The global fault tolerant optimization of a wind farm in a distributed framework is of interest for scalability, maintainability and reliability and therefore will have great economical impact.
So far the problem of wind farm control only considered the problem of nominal performance optimization. The challenge is still open to develop algorithms to accurately update and reconfigure the distributed controller when the wind profile is changing, and/or when the performance of individual components on wind turbines degrades, fault tolerant control. For example when a turbine fails this has to be detected, communicated towards the other turbines and their controller should be reconfigured to optimize the performance of the wind farm (global performance).
Objectives of this project are:
This PhD project is part of the Far and Large OffshoreWind (FLOW) innovation program.
- Set-up of a state-of-the-art virtual simulation environment that models the dynamics of the wind farm on a time-scale appropriate for real-time control.
- Optimize the energy production by the development of a distributed controller, taking the global (spatial) changes in the wind profile into account.
- Development of novel system identification techniques and adaptive control strategies for nonlinear/distributed systems. This enables taking into account changes in the dynamics of the wind turbine system due to wear and tear or changing wind profiles, and adapting the controller accordingly: the data-driven control methodology.
- The methodology is evaluated on a number of challenging load cases and fault scenarios. In which the effectiveness of the proposed approach is presented and computational aspects are investigated.