Title :
Self-tuning disturbance feedforward control with drift prevention for air mount systems
Author :
Michiel A. Beijen;Marcel F. Heertjes;Hans Butler
Author_Institution :
Department of Mechanical Engineering, Group of Control Systems Technology, Eindhoven University of Technology, 5600 MB Eindhoven, The Netherlands
Abstract :
A MIMO disturbance feedforward control strategy is presented to isolate an industrial active vibration isolation system with air mounts from broadband floor vibrations. The feedforward controller compensates for the static damping and stiffness of the air mount suspension, leading to significant improvement of the vibration isolation performance. At low frequencies the controller gain is limited using higher-order weak integrators to prevent drift and actuator saturation. To minimize performance limitations due to model uncertainties, the MIMO feedforward controller is implemented as an IIR filter with fixed poles and self-tuning zeros, having the ability to fine-tune the parameters online using a combination of filtered-ϵ least mean squares (FϵLMS) optimization and residual noise shaping. An experimental validation on a full-scale vibration isolation system with air mounts shows performance improvements up to 30 dB for frequencies between 30 and 80 Hz.
Keywords :
"Feedforward neural networks","Vibrations","Actuators","Frequency control","Suspensions","Isolators","Least squares approximations"
Conference_Titel :
Control Applications (CCA), 2015 IEEE Conference on
DOI :
10.1109/CCA.2015.7320662