Noise Steering: Exploiting Wake Steering and Acoustic Directivity for Noise-Constrained Wind Farm Control via Reinforcement Learning

A model-free reinforcement learning strategy jointly optimizes turbine yaw angles and power setpoints to maximize wind farm output under local noise constraints. By coupling a low-fidelity acoustic model with the FLORIS simulator, the agent learns to redirect both wakes and noise directivity away from protected regions.