Wind Farm Power Optimization Using Bayesian Adaptive Direct Search for Active Pitch Control

The rapid development of the wind energy sector, particularly in China, has led to significant advancements towards carbon neutrality and renewable energy utilization. However, optimizing wind farm control strategies to enhance power output and economic efficiency remains a critical challenge, particularly in mitigating wake effects between turbines. This study focuses on addressing these challenges through the implementation of active pitch control strategies. The Bayesian-Adaptive Direct Search (BADS) algorithm, which combines the global search capabilities of Bayesian Optimization with the local refinement strengths of Mesh Adaptive Direct Search, is utilized for real-time optimization. This method effectively navigates the high-dimensional parameter space, accommodating non-smooth pitch angle settings and complex thrust coefficient curves. Open-FAST, from the National Renewable Energy Laboratory (NREL), serves as the primary simulation tool. Through simulations conducted on three NREL 5MW turbines, the proposed strategy demonstrated a 4.7% increase in total power output by effectively reducing wake effects and enhancing energy utilization. This research not only highlights the inadequacies of current parameter control and optimization methods but also presents a viable solution for improving wind farm performance.