AAU Student Projects - visit Aalborg University's student projects portal
A master's thesis from Aalborg University
Book cover


Model Predictive LIDAR Control of Wind Turbines for Load Mitigation

Authors

;

Term

10. term

Publication year

2011

Submitted on

Pages

120

Abstract

Vindmøller udsættes for skiftende vind, som giver mekaniske belastninger. Denne afhandling udvikler en modelprædiktiv styring (MPC), en metode der forudsiger fremtidig adfærd for at optimere styringen, og bruger LIDAR (laserbaseret måling af vind) til at forudse den vind, der er på vej, så belastninger kan reduceres. Til simulering udledes en ikke-lineær model af en vindmølle, som valideres ved at kontrollere fysiske egenskaber og ved at sammenligne med en model med tilsvarende detaljeringsgrad samt en mere kompleks aeroelastisk model i programmet LACflex. For at kunne anvende MPC’en linearisers den ikke-lineære model. Der udvikles også en LIDAR-model, der leverer gennemsnitlige vindhastigheder fra vindfeltet i LACflex. MPC’en implementeres og testes i MATLAB og LACflex. I MATLAB sammenlignes den med en referencecontroller: en standard PI-controller (proportional-integral) med ekstra dæmpning på tårn og drivlinje. I LACflex testes den på en mere kompleks møllemodel og sammenlignes med LACflex’ indbyggede PI-controller. Simulationerne viser, at ved at inkludere LIDAR-målinger i MPC kan belastninger reduceres, mens effekt opretholdes.

Wind turbines face changing winds that create mechanical loads. This thesis designs a model predictive controller (MPC)—a control method that predicts future behavior to optimize actions—and uses LIDAR (laser-based wind sensing) to anticipate incoming wind and reduce loads. For simulation, a nonlinear wind turbine model is derived and validated by checking physical characteristics and by comparing it with a model of similar detail and a more complex aeroelastic model in the LACflex program. To make it usable for the controller, the nonlinear model is linearized. A LIDAR model is also developed to provide averaged wind speed measurements from the wind field defined in LACflex. The MPC is implemented and tested in MATLAB and LACflex. In MATLAB, it is compared with a benchmark controller: a standard proportional-integral (PI) controller with additional damping on the tower and drive train. In LACflex, it is tested on a more complex turbine model and compared with LACflex’s built-in PI controller. Simulation results show that including LIDAR measurements in the MPC can mitigate loads while maintaining power.

[This abstract was generated with the help of AI]