Data-Driven Approach for Wind Farm Control: Toward an Alternative to FLORIS

In this paper, we introduce a data-driven approach to wind farm control, offering an alternative to the FLORIS wind farm simulator. Our method estimates the power of a wind farm and determines the optimal yaw angle to maximize power generation. Initially, we develop a power estimation neural network using data from FLORIS for power estimation and validate its accuracy and reliability. Subsequently, this power estimation neural network is employed to determine the optimal yaw angle for maximum power production. The efficacy of this yaw decision neural network is verified through various performance metrics. We then present dynamic simulations by integrating the yaw decision neural network, constructed through our data-driven approach, with a dynamic wind farm simulator. We believe this addresses the limitations of FLORIS, a steady-state simulator. Our results demonstrate the effectiveness of the proposed yaw decision neural network in dynamic environments, underscoring the potential of a data-driven approach to overcome the challenges posed by the steady-state wind farm simulator. This study offers innovative solutions for the efficient control and optimization of wind farm.