aiqbench analysis
This study applies offline reinforcement learning to optimize wind farm power generation by managing yaw control under dynamic wind conditions. By leveraging precollected datasets, the proposed MTD3-BC algorithm achieves significant efficiency gains and mitigates wake effects without requiring extensive real-time simulator interactions or complex wake models. The experimental validation in a wind tunnel marks a notable advancement in using data-driven RL for renewable energy infrastructure control.
Offline Reinforcement Learning for Wind Farm Control: A Wind Tunnel Study under Dynamic Wind Directions
arXiv:2609.12905v1 Announce Type: new Abstract: This paper addresses the wind farm power maximization problem in the presence of wind direction changes. Specifically, a model-free Modified Twin Delayed Deep Deterministic Policy Gradient with Behavior Cloning (MTD3-BC) algorithm is proposed to tackle this task through yaw control under varying wind direction conditions. MTD3-BC is an offline reinforcement learning (RL) algorithm that aims to infer good behavior from only a precollected offline dataset. Additionally, to ensure smooth and moderate yaw adjustments, a new action consistency term is introduced into the policy optimization objective. Unlike online RL methods, MTD3-BC does not require extensive interactions with a wind farm simulator during training, significantly reducing computational costs and training time. A wind tunnel experiment is conducted to validate the effectiveness of the algorithm under varying wind directions. The results demonstrate that MTD3-BC successfully mitigates wake effects, delivering farm-level power gains of approximately 10\% over the baseline greedy strategy and performance on par with a data-calibrated model-based wake-steering benchmark, while requiring no wake model and only a small fraction of the training cost of online RL. To our knowledge, this is the first time an offline RL wind farm control policy has been validated and demonstrated experimentally.
Abstract:This paper addresses the wind farm power maximization problem in the presence of wind direction changes. Specifically, a model-free Modified Twin Delayed Deep Deterministic Policy Gradient with Behavior Cloning (MTD3-BC) algorithm is proposed to tackle this task through yaw control under varying wind direction conditions. MTD3-BC is an offline reinforcement learning (RL) algorithm that aims to infer good behavior from only a precollected offline dataset. Additionally, to ensure smooth and moderate yaw adjustments, a new action consistency term is introduced into the policy optimization objective. Unlike online RL methods, MTD3-BC does not require extensive interactions with a wind farm simulator during training, significantly reducing computational costs and training time. A wind tunnel experiment is conducted to validate the effectiveness of the algorithm under varying wind directions. The results demonstrate that MTD3-BC successfully mitigates wake effects, delivering farm-level power gains of approximately 10\% over the baseline greedy strategy and performance on par with a data-calibrated model-based wake-steering benchmark, while requiring no wake model and only a small fraction of the training cost of online RL. To our knowledge, this is the first time an offline RL wind farm control policy has been validated and demonstrated experimentally.
Machine Learning (cs.LG)
arXiv:2609.12905 [cs.LG]
(or arXiv:2609.12905v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.12905 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 289 words
Related coverage
High-Altitude Balloon Station-Keeping with First Order Model Predictive Control
arXiv · Sep 4, 2026
Sampling-Based MuJoCo MPC for Double-Pendulum Sway Suppression on a Shipboard Crane
arXiv · Sep 25, 2026
Offline Reinforcement Learning for Plasma Control in Nuclear Fusion: Codebase and Benchmark
arXiv · Jun 9, 2026
Individual Control Barrier Functions-Guided Diffusion Model for Safe Offline Multi-Agent Reinforcement Learning
arXiv · Jun 12, 2026