Iterative learning control for load management in wind turbines with smart rotor blades

Abstract

Control of aerodynamic loads is a crucial issue in keeping wind energy economically competitive with traditional energy sources. Loads on wind turbine blades can be managed through collective and individual pitch control, however recently there has been significant research on application of active flow control devices which can alter the flow locally, spanwise on the blade. This work investigates the use of iterative learning control for load control in wind turbines with smart rotor blades providing a significant extention of the previous research on model-free design and a substantial contribution on the model-based approach. Iterative learning control is capable of rejecting periodic disturbances in systems performing repetitive tasks and in this particular application it is used to modify the blade section aerodynamics such that the fluctuations in load due to periodic disturbances on the blades are minimized. At first, a computational fluid dynamics flow model with a basic structure iterative learning control law is used, where the controller’s gains are chosen without the use of a model of the dynamics akin to auto-tuning design. Model-free design demonstrates the potential of this algorithm for wind turbines control but is limited in what it can deliver, especially as testing is computationally ineffective as it requires running the full computational fluid dynamics simulation each time. Subsequently, model based design is considered where a Proper Orthogonal Decomposition based reduced order model is constructed and used to design and test the norm optimal iterative learning control scheme. Construction of a reduced order model requires running the full computational fluid dynamics simulation only once and various controllers can be designed and evaluated using this low-dimensional model. The performance of designed controllers is evaluated in simulation for the state-space model and in full computational fluid dynamics test and the results show that aerodynamic load on the blade can be successfully controlled by iterative learning control.

More information

Identifiers

ORCID for Weronika Natalia Nowicka: orcid.org/0000-0002-7049-1162

ORCID for Bing Chu: orcid.org/0000-0002-2711-8717

Catalogue record

Export record