Repetitive Control of Individual Pitch to Reduce Wake Effect on Wind Turbines
Authors
Nørby, Kåre Engell ; Pedersen, Anders
Term
10. term
Publication year
2012
Submitted on
2012-05-31
Pages
166
Abstract
Efterhånden som vindmøller bliver større, med rotordiametre over 100 m, fejer vingerne gennem et vindfelt med varierende forhold som våger, vindskæring (ændringer i vindhastighed med højden) og tårnskygge. Det giver betydelige, periodiske belastninger på konstruktionen. Denne afhandling udvikler en lifted repetitive controller for at reducere disse belastninger ved hjælp af individuel pitchregulering, hvor hver vinges vinkel justeres uafhængigt. Til formålet er der opbygget en dynamisk model af en vindmølle, der omfatter en aerodynamisk model, en mekanisk model, en strukturel model og en model af pitchsystemet. Modellen blev lineariseret for at forenkle styringsdesignet og valideret mod simulationskoden FAST. På basis af modellen blev der formuleret en lifted systembeskrivelse, og en reduceret output-feedback blev anvendt, så et LQR-design (Linear Quadratic Regulator) kunne beregne regulatorforstærkningerne. I en accepttest blev den nye controller sammenlignet med controlleren fra FAST, som var implementeret i Matlab. Den designede controller bestod ikke testen, selv om tårn- og vingebøjning blev reduceret. Det antages, at dette skyldes en uoverensstemmelse mellem modellen og implementeringen i Matlab.
As wind turbines grow to rotor diameters above 100 m, each rotation sweeps through wind with varying conditions—such as wakes, wind shear (changes in wind speed with height), and tower shadow—that create significant, periodic structural loads. This thesis develops a lifted repetitive controller to reduce these loads using individual pitch control, which adjusts each blade’s angle independently. To support the design and testing, a dynamic wind turbine model was built with aerodynamic, mechanical, structural, and pitch-system components. The model was linearized to simplify control design and validated against the simulation code FAST. Based on this model, a lifted system description was formulated, and a reduced output-feedback approach was used so that Linear Quadratic Regulator (LQR) design could determine the controller gains. In an acceptance test, the new controller was compared with the FAST controller implemented in Matlab. The designed controller did not pass the test, even though tower and blade deflections decreased. It is assumed that this is due to a mismatch between the model and the Matlab implementation.
[This abstract was generated with the help of AI]
Keywords
Documents