MS#10.4 Turbine control benchmarking, algorithms, and learning

S. MULDERS¹, D. ZALKIND², T. GYBEL HOVGAARD³, J. DELEURAN GRUNNET³
¹ Delft University of Technology|² National Renewable Energy Laboratory (NREL)|³ Shanghai Electric Wind Power Group (SEWPG)

Emerging technologies and special sessions

This mini-symposium focuses on key research areas identified as crucial by academics and the industry. These topics include benchmarking frameworks for evaluating controller innovations in relevant load cases, novel control algorithms for flexible wind turbines, and learning algorithms that recalibrate internal control models in real time to reflect the current state of the turbine. While these are the core themes, the session also welcomes research contributions on other aspects of onshore and offshore wind turbine control.

To better evaluate the impact of research control innovations and understand their industrial relevance, the organizers are collaborating to create a benchmarking procedure that will automate performance evaluations across a range of operating conditions. The organizers plan to share this procedure during this mini-symposium, begin a dialog on the process, encourage adoption, and pose control challenges for the community of wind turbine researchers to solve.

The session is followed up with research contributions in advanced control and learning algorithms for wind turbines. Many state-of-the-art industrial controllers assume rigid turbine structures, which limits their ability to meet the more complex demands of modern, flexible turbines. Novel control strategies are needed to account for the aerodynamic impacts of turbine flexibility, the interaction with substructures, and the overall operational strategy required for optimized performance.

Furthermore, emphasis is placed on learning control algorithms, as they hold significant potential in addressing the limitations of current industrial controllers. Industrial wind turbine controllers often rely heavily on the accuracy of the internal model parameters. However, these modeled parameters frequently deviate significantly from the actual aerodynamic properties of the turbine right from the start. Over time, this gap widens due to factors like blade erosion, ice accumulation, and debris buildup. Therefore, to sustain control performance, developing learning algorithms that can recalibrate these internal models is of critical importance.

This mini-symposium is a collaborative endeavor between academic research groups and industrial control experts. With this symposium, we aim to foster stronger partnerships between the two fields. We anticipate a platform for disseminating current research, exchanging novel research ideas, networking, and conducting direct discussions during the symposium, ensuring that everyone is an integral part of this collective effort to push the boundaries of wind energy technology.