Decentralized Model-Free Wind Farm Control via Discrete Adaptive Filtering Methods

The aim of this paper is to present a decentralized model-free approach to wind farm power optimization with limited information sharing among the turbines. Two decentralized discrete adaptive filtering algorithms are proposed to optimize a wind farm's total power output without utilizing the wind farm power generation model, and with only limited information sharing among neighbor turbines. Convergence results of the proposed algorithms are presented. The proposed algorithms are further extended to track the time-varying environment. Simulation results show that when turbines in the wind farm employ the proposed decentralized algorithms, the total power output of the turbines quickly converges to or very close to the optimal total power generation. The ability to maximize the power output in time-varying environment is also demonstrated. Finally, the proposed algorithms are demonstrated to be robust in realistic conditions, where large magnitude environmental disturbances are considered.