model reference adaptive control system using frequency domain performance specifications
abstract
mras (model reference adaptive system) to achieve frequency domain performance specifications
with different transfer function identification techniques are experimentally studied.
standard open loop and closed loop recursive least squares (rls) system identification
techniques are implemented along with methods based on bandpass filters and steady-state
kalman filter. the behavior of the mras when combined with the different system identification
techniques is discussed.
based on experimental work performed on mras using open loop system identification,
it was found that the system can achieve performance specifications in all process conditions.
it was noticed that the dc bias changes in process input when process changes, delays
the convergence of the system identification in open loop system identification. closed
loop system identification was found inadequate for use in mras. the system identification
based on bandpass filter approach was found slow in convergence and so inefficient
for use in mras. the use of kalman filter for system identification resulted in noisy gains
on adaptive pi controller. based on the results obtained, it is concluded that mras to
achieve frequency domain performance specifications is practical and capable of being used
in industrial process control when coupled with open loop identification based on frequency
domain concepts. the proposed approach provides a better method for maintaining performance
robustness while guaranteeing stability margins in adaptive control than that
obtained from time domain adaptive control.
collections
- retrospective theses [1604]