Condition Monitoring of Hydraulic Actuator in Wind Turbine Pitch Systems: Investigation and EKF Implementation
Author
Andersen, Niklas Enoch
Term
4. term
Education
Publication year
2023
Submitted on
2023-06-02
Pages
46
Abstract
This thesis investigates condition monitoring of internal leakage in the hydraulic actuator of a wind turbine pitch system. A brief state-of-the-art review of relevant methods is conducted, and an extended Kalman filter (EKF)-based approach is selected for detailed study. A nonlinear model of the system is developed and validated against experimental data from a pitch-system test setup, and deemed sufficiently accurate for comparison purposes. Based on this, a linear state-space model is derived for the EKF. The filter is designed to estimate system states and internal leakage and is evaluated against the nonlinear model in two test campaigns. The initial test fails due to diverging covariances traced to simplifications introduced during linearization; after accounting for these in the prediction step, performance improves markedly and the second test succeeds. Results show accurate pressure estimation, while blade angle and angular velocity exhibit some deviations and noise. The leakage estimate is usable but relatively noisy, which can obscure interpretation. The study concludes that the EKF is a promising tool for condition monitoring of internal leakage in hydraulic pitch actuators, but it requires further tuning, broader testing across operating conditions, and supporting algorithms to enhance robustness and predict when leakage becomes critical.
Dette speciale undersøger tilstandsovervågning af interne lækager i den hydrauliske aktuator i et vindmølles vinkelstyringssystem (pitch). Først gennemføres en kort state-of-the-art-undersøgelse af relevante metoder, hvorefter en udvidet Kalman-filter (EKF)-baseret tilgang udvælges til nærmere studie. Der udvikles en ulineær model af systemet, som valideres mod eksperimentelle data fra et forsøgsopstillet pitch-system og vurderes tilstrækkeligt nøjagtig til sammenligninger. På baggrund heraf udledes en lineær tilstandsrum-model til brug i EKF’et. Filteret designes til at estimere systemets tilstande og den interne lækage og sammenlignes med den ulineære model i to testforløb. Den første test mislykkes pga. divergerende kovarianser, hvilket spores til forenklinger indført ved linearisering; efter at tage højde for disse i forudsigelsen forbedres performance markant, og anden test lykkes. Resultaterne viser god estimering af tryk, mens vinkel og vinkelhastighed har visse afvigelser og støj. Lækageestimatet er brugbart, men relativt støjfyldt, hvilket kan sløre fortolkningen. Konklusionen er, at EKF’et er lovende til tilstandsovervågning af intern lækage i hydrauliske pitch-aktuatorer, men kræver yderligere tuning, mere omfattende afprøvning under flere driftsforhold og supplerende algoritmer til mere robust overvågning og forudsigelse af kritiske lækageniveauer.
[This apstract has been generated with the help of AI directly from the project full text]