Researchers from Imperial College London, including Andrew Mole, Max Weissenbacher, Georgios Rigas, and Sylvain Laizet, have developed a new approach to optimize wind farm power production using reinforcement learning (RL). Their work, published in the journal Nature Energy, focuses on improving the overall efficiency of wind farms by coordinating the control of individual turbines.
Traditionally, wind farms operate each turbine independently to maximize individual power output. However, this approach can lead to inefficiencies due to the wake effects created by upstream turbines, which can reduce the power output of downstream turbines. The researchers aimed to address this issue by developing a dynamic, closed-loop control system that coordinates the operation of all turbines in a wind farm to minimize wake effects and maximize overall power output.
The key innovation in this research is the integration of a reinforcement learning controller with high-fidelity large-eddy simulation (LES). This allows the control system to respond in real-time to atmospheric turbulence and other dynamic conditions, enabling more effective coordination of turbine operations. The researchers found that their RL controller achieved a 4.30% increase in wind farm power output compared to baseline operation, nearly doubling the 2.19% gain from static optimal yaw control obtained through Bayesian optimization.
The practical applications of this research for the energy sector are significant. By improving the efficiency of wind farms, this technology can help accelerate the deployment of renewable energy and contribute to achieving net-zero targets. The dynamic, flow-responsive control approach developed by the researchers could be particularly valuable in regions with complex atmospheric conditions, where traditional control strategies may be less effective.
In summary, the researchers from Imperial College London have demonstrated that reinforcement learning can be used to optimize wind farm power production by enabling closed-loop collaborative control. This approach has the potential to improve the efficiency of wind farms and contribute to the broader goal of transitioning to a more sustainable energy system.
This article is based on research available at arXiv.