Adaptive Command Filtering to Compensate for the Oscillatory Motion in Elastic Systems

by David P. Magee

Advisor: Wayne J. Book

Current domestic and space applications for active control of unwanted vibration involve systems whose fundamental frequencies are within the desired bandwidth of operation. To permit operational frequencies above those of the system, the position controller must be designed with information about the oscillatory behavior of the system and make proper adjustments to achieve acceptable system performance. The proposed research suggests an adaptive filtering method to compensate for the oscillatory behavior associated with the compliant subsystem and a feedback controller that will yield overall system motion that is free of vibration. For linear time invariant systems, the proposed filtering algorithm should reduce in principle to the input shaping method by Singer and Seering.

The design of an effective filtering algorithm will require an accurate representation of the oscillatory motion of the system. In complex systems, the motion can be very difficult to represent mathematically for all possible configurations. To overcome these situations, a system identification algorithm will be incorporated into the filtering algorithm to adjust filter parameters that depend upon the system's configuration. If the filter parameters do change, the algorithm must properly adapt to maintain a smooth output throughout the transition so that vibrations are not excited in the system.

Once an acceptable filtering procedure is developed, it will be combined with a feedback controller to position the system. Theoretical studies will investigate the change in position of the system due to the filtering method as well as the overall stability of the control scheme. The combined adaptive filtering and feedback control strategy will be verified with an experimental two link flexible manipulator with two degrees of actuation named RALF and an experimental flexible manipulator with three degrees of actuation named BERTHA, both located at the Georgia Institute of Technology. The proposed research should produce a robust vibration control scheme that can be used with various autonomous as well as telemanipulator position trajectories.


David P. Magee
dpm@euler.marc.gatech.edu
Last Revised: November 27, 1995