A reinforcement‐learning approach for individual pitch control
Abstract Individual pitch control has shown great capability of alleviating the oscillating loads experienced by wind turbine blades due to wind shear, atmospheric turbulence, yaw misalignment, or wake impingement. This work presents a novel controller structure that relies on the separation of low‐...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2022-08-01
|
| Series: | Wind Energy |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/we.2734 |
| _version_ | 1811291907880386560 |
|---|---|
| author | Marion Coquelet Laurent Bricteux Maud Moens Philippe Chatelain |
| author_facet | Marion Coquelet Laurent Bricteux Maud Moens Philippe Chatelain |
| author_sort | Marion Coquelet |
| collection | DOAJ |
| description | Abstract Individual pitch control has shown great capability of alleviating the oscillating loads experienced by wind turbine blades due to wind shear, atmospheric turbulence, yaw misalignment, or wake impingement. This work presents a novel controller structure that relies on the separation of low‐level control tasks and high‐level ones. It is based on a neural network that modulates basic periodic pitch angle signals. This neural network is trained with reinforcement learning, a trial and error way of acquiring skills, in a low‐fidelity environment exempt from turbulence. The trained controller is further deployed in large eddy simulations to assess its performances in turbulent and waked flows. Results show that the method enables the neural network to learn how to reduce fatigue loads and to exploit that knowledge to complex turbulent flows. When compared to a state‐of‐the‐art individual pitch controller, the one introduced here presents similar load alleviation capacities at reasonable turbulence intensity levels, while displaying very smooth pitching commands by nature. |
| first_indexed | 2024-04-13T04:36:26Z |
| format | Article |
| id | doaj.art-f4865a4e6eff49fbb53c37c9a9d63e70 |
| institution | Directory Open Access Journal |
| issn | 1095-4244 1099-1824 |
| language | English |
| last_indexed | 2024-04-13T04:36:26Z |
| publishDate | 2022-08-01 |
| publisher | Wiley |
| record_format | Article |
| series | Wind Energy |
| spelling | doaj.art-f4865a4e6eff49fbb53c37c9a9d63e702022-12-22T03:02:09ZengWileyWind Energy1095-42441099-18242022-08-012581343136210.1002/we.2734A reinforcement‐learning approach for individual pitch controlMarion Coquelet0Laurent Bricteux1Maud Moens2Philippe Chatelain3Institute of Mechanics, Materials and Civil Engineering Université catholique de Louvain Louvain‐la‐Neuve BelgiumFluids‐Machines Department Université de Mons Mons BelgiumInstitute of Mechanics, Materials and Civil Engineering Université catholique de Louvain Louvain‐la‐Neuve BelgiumInstitute of Mechanics, Materials and Civil Engineering Université catholique de Louvain Louvain‐la‐Neuve BelgiumAbstract Individual pitch control has shown great capability of alleviating the oscillating loads experienced by wind turbine blades due to wind shear, atmospheric turbulence, yaw misalignment, or wake impingement. This work presents a novel controller structure that relies on the separation of low‐level control tasks and high‐level ones. It is based on a neural network that modulates basic periodic pitch angle signals. This neural network is trained with reinforcement learning, a trial and error way of acquiring skills, in a low‐fidelity environment exempt from turbulence. The trained controller is further deployed in large eddy simulations to assess its performances in turbulent and waked flows. Results show that the method enables the neural network to learn how to reduce fatigue loads and to exploit that knowledge to complex turbulent flows. When compared to a state‐of‐the‐art individual pitch controller, the one introduced here presents similar load alleviation capacities at reasonable turbulence intensity levels, while displaying very smooth pitching commands by nature.https://doi.org/10.1002/we.2734individual pitch controllarge eddy simulationload alleviationreinforcement learning |
| spellingShingle | Marion Coquelet Laurent Bricteux Maud Moens Philippe Chatelain A reinforcement‐learning approach for individual pitch control Wind Energy individual pitch control large eddy simulation load alleviation reinforcement learning |
| title | A reinforcement‐learning approach for individual pitch control |
| title_full | A reinforcement‐learning approach for individual pitch control |
| title_fullStr | A reinforcement‐learning approach for individual pitch control |
| title_full_unstemmed | A reinforcement‐learning approach for individual pitch control |
| title_short | A reinforcement‐learning approach for individual pitch control |
| title_sort | reinforcement learning approach for individual pitch control |
| topic | individual pitch control large eddy simulation load alleviation reinforcement learning |
| url | https://doi.org/10.1002/we.2734 |
| work_keys_str_mv | AT marioncoquelet areinforcementlearningapproachforindividualpitchcontrol AT laurentbricteux areinforcementlearningapproachforindividualpitchcontrol AT maudmoens areinforcementlearningapproachforindividualpitchcontrol AT philippechatelain areinforcementlearningapproachforindividualpitchcontrol AT marioncoquelet reinforcementlearningapproachforindividualpitchcontrol AT laurentbricteux reinforcementlearningapproachforindividualpitchcontrol AT maudmoens reinforcementlearningapproachforindividualpitchcontrol AT philippechatelain reinforcementlearningapproachforindividualpitchcontrol |