02 November 2023
“Neural Networks Enhance Wind Inflow Observation”
A new study from TUM could improve the performance of wind turbines and wind farms
A recent research paper introduces a significant advancement in the field of wind energy technology. The study employs feed-forward neural networks to estimate wind inflow in real-time during wind turbine operation, replacing a previous piecewise-linear model.
Simplifying tuning and usage, the system decouples wind parameters, enabling independent observation. To validate the concept, simulations and data from a 3.5 MW wind turbine, along with nearby met mast measurements, were employed for training and verification.
The findings show that using neural networks gives us better estimates for wind patterns and turbine alignment, especially when the wind is not too strong. This new method could help make wind turbines and wind farms work better, making them more efficient and effective.
The paper titled “Wind inflow observation from load harmonics via neural networks: A simulation and field study” was published by the Technical University of Munich, one of MERIDIONAL’s partners.
Authors: Kwang-Ho Kim, Marta Bertelè, Carlo L. Bottasso
Read the paper here!
Picture from Freepik.