Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Adaptive control of the PMSG based drive train for active vibration damping

Presentation held at COWEC 2013, June 18-19, Berlin, Germany
 
: Chen, L.; Wenske, J.

:
Präsentation urn:nbn:de:0011-n-3371922 (603 KByte PDF)
MD5 Fingerprint: e9222e096a143dd034ced1e6f3187fa3
Erstellt am: 28.4.2015


2013, 19 Folien
Wind Power Engineering Community (COWEC Conference) <2013, Berlin>
Englisch
Vortrag, Elektronische Publikation
Fraunhofer IWES ()
adaptive control; active torsional vibration damping; drive train

Abstract
Wind turbines with permanent magnet synchronous generators (PMSG) as direct drive (DD) or geared drive train design with inherent full converter grid connection are now becoming preferred, especially in multi-megawatt range. PMSG based wind turbines have higher efficiencies, less mechanical components in the DD drive trains variants, better electrical power quality, etc.. But torsional oscillations within the drive train between the rotor and the air gap of the generator remain an issue, especially if the wind speed suddenly changes and the wind distribution on the rotor is very inhomogeneous or characterized by high local fluctuations. The conventional PI/PID controllers have good performance on the speed control/power curve control, but are not able to minimize the torsional oscillations within the drive train. The purpose of this paper is to present an adaptive control for a PMSG-based drive train. The main control objective is still to keeping the rotor rotating with an optimal tip speed ratio. But the torsional oscillations can be suppressed even with the variable wind speed. The controller is designed with the Backstepping technique. This approach treats the state variables as virtual control inputs. The design process is recursive and the controller can be step by step straightforwardly derived. The stability can be proved by the Lyapunov Theory. With the Backstepping controller the speed error will exponentially converge to zero. The backstepping controller is designed with the two-mass model of the drive train. In the practice the shaft torque and aerodynamic torque cannot be accurately measured. Therefore, the Luenberger observer is applied to estimate the shaft torque. The aerodynamic torque is estimated online by the parameter adaption law, which can be easily integrated in the Backstepping algorithm. Thus the oscillation caused by the change of the wind can be suppressed. For this controller design only the measurements of the rotor speed, generator speed and generator power are necessary. The adaptive Backstepping controller is tested with simulations based on the model of a 2MW direct-drive PMSG wind turbine in Matlab/Simulink. This controller is compared with the conventional PI controller. Simulation results show that both controllers can achieve very good speed tracking. But the adaptive controller suppresses the torsional vibration significantly without loss of energy yield.

: http://publica.fraunhofer.de/dokumente/N-337192.html