Active Vibration Control of Wind Turbine Generator Drive Train
Authors
Janek, Jonas Amager ; Christensen, Nicolai Steensgaard ; Pedersen, Patrick Nedergaard
Term
4. term
Education
Publication year
2026
Submitted on
2026-05-28
Abstract
This thesis investigates active vibration control of a wind turbine generator drivetrain by manipulating generator torque to suppress gearbox vibrations. Focusing on a two-stage planetary gearbox, vibration measurements are analyzed using FFT to identify the gear mesh frequency as the dominant, speed-dependent disturbance. A data-driven primary vibration model combines a Fourier representation with Gaussian process regression to predict vibration amplitudes as a function of speed and torque, while a secondary-path model is identified from measurements with sinusoidal torque inputs to capture how control actions affect vibrations; the path behaves like a torque-dependent notch filter. Building on these models, an adaptive Filtered-x LMS controller generates a compensating sinusoidal torque targeting the first-order gear mesh component. Simulation results using measurement data provided by Vestas and varying operating conditions show effective attenuation across speeds and loads with stable convergence, although validation is limited to simulation due to restricted hardware access. Overall, the study indicates that generator-torque-based adaptive control is a promising approach to reduce gearbox vibrations and support drivetrain reliability under changing wind conditions.
Dette speciale undersøger aktiv vibrationskontrol af en vindmøllegenerators drivtog ved at manipulere generatormomentet for at dæmpe vibrationer i gearkassen. Fokus er på en to-trins planetgear, hvor vibrationsmålinger analyseres med FFT for at identificere tandindgrebsfrekvensen som den dominerende, hastighedsafhængige forstyrrelse. En data-drevet primær vibrationsmodel kombinerer en Fourier-beskrivelse med Gaussian Process Regression til at forudsige vibrationsamplituder som funktion af hastighed og moment, mens en sekundær-sti-model identificeres ud fra målinger med sinusformede momentinput for at beskrive, hvordan kontrolhandlinger påvirker vibrationerne; stien opfører sig som et momentafhængigt notch-filter. Med udgangspunkt i disse modeller implementeres en adaptiv Filtered-x LMS-regulator, der genererer et kompenserende sinusformet moment for målrettet at reducere førsteordens tandindgrebskomponenten. Simulationsresultater med måledata leveret af Vestas og varierende driftsbetingelser viser effektiv dæmpning på tværs af hastigheder og laster med stabil konvergens, om end valideringen er begrænset til simulation på grund af begrænset hardwareadgang. Samlet peger arbejdet på, at adaptiv styring via generatormoment er en lovende metode til at reducere gearkassevibrationer og understøtte højere pålidelighed i drivtoget under skiftende vindforhold.
[This apstract has been generated with the help of AI directly from the project full text]