Biomimetic individual pitch control for load alleviation

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 misalignement or wake impingement. This work presents a bio-inspired structure for individual pitch control where neural oscillators produce basic rhythmic patterns of the pitch angles, while a deep neural network modulates them according to the environmental conditions. This mimics, respectively, the central patterns generators present in the spinal chord of animals and their cortex. The mimicry further applies to the neural network as it is trained with reinforcement learning, a method inspired by the trial and error way of animal learning. Large eddy simulations of the reference NREL 5MW wind turbine using this biomimetic controller show that the neural network learns how to reduce fatigue loads by producing smooth pitching commands.

Handling Individual Pitch Control within an Actuator Disk framework: verification against the Actuator Line method and application to wake interaction problems

“…This first study showed promising results and gave us confidence in deploying IPC strategies within an AD framework. However, this approach should be first verified in more complex environments (with a realistic Atmospheric Boundary Layer (ABL) representation, in a wind farm with more than two machines, etc) and with other IPC schemes [29].…”

A reinforcement‐learning approach for individual pitch control

“…The NN input layer is thus made of n S þ 4 neurons. Three hidden layers, comprising *In the previous investigations by the authors, 31 the state vector did not include the sector-effective load oscillation ϵ, but rather a sector-effective infinite upstream velocity. The latter was estimated from the blade loads in the fashion of Bottasso et al 40 The direct use of blade loads at a sector level was expected to reduce the uncertainties and delays associated to estimations.…”

“…The envisioned paradigm should enable an improved management of the aerodynamic loads that will be encountered in future large rotors. To do so, this work builds on previous investigations by the authors 31 to go further in terms of controller architecture and performance assessment. RL is sometimes referred to as a bio‐inspired control mechanism, as it mimics an animal's way of learning by experience 23 .…”

Assessment of an actuator disk‐based approach for the prediction of fatigue loads in a perspective of wind farm‐scale application

“…It is thus of a crucial importance to deploy efforts in the prediction of the fatigue damage a wind turbine will suffer during its operation, either to optimally design the wind farm or to develop load alleviation strategies that can be used during the wind farm operation. Those strategies encompass load alleviation at a wind turbine level 3–5 but also schemes aiming at ensuring a more uniform and potentially reduced wear and tear of the rotors at a wind farm scale 6,7 …”

“…For some wind turbine subassemblies, the downtime associated with repair may be significant, mainly in the offshore environment: This decreases the wind turbine reliability, directly impacting the Levelized Cost of Energy, as demonstrated in Dao et al 2 It is thus of a crucial importance to deploy efforts in the prediction of the fatigue damage a wind turbine will suffer during its operation, either to optimally design the wind farm or to develop load alleviation strategies that can be used during the wind farm operation. Those strategies encompass load alleviation at a wind turbine level [3][4][5] but also schemes aiming at ensuring a more uniform and potentially reduced wear and tear of the rotors at a wind farm scale. 6,7 As the development of control strategies or the potential arrangement of a future wind farm often relies on numerical studies, it is essential that the simulation tools provide accurate blade/structural loads and fatigue estimates.…”