This is not specifically a python-control question; I am taking advantage
of the community of controls minds.
I have a continuous system (a robot with one flexible link - basically a
slewing beam) that uses accelerometer feedback for vibration suppression
(admittedly, accel. feedback is a bit risky). I have designed a control
system using a continuous approach and now I want to compare that to a
state-space design based on a reduced order model. The accelerometer makes
the system susceptible to higher modes being driven unstable. In my
continuous design, I solve this by making sure the accelerometer feedback
loop has a lowpass filter.
My question is this: is there a way to do an LQG design that requires the
final design to include a lowpass filter? I want the filter design to also
be some how optimized as part of the LQG design (i.e. I don't want to pick
the filter corner frequency myself first).
As a hack/work around, I was thinking of setting it up as a numeric
optimization problem where the lowpass corner frequency is the thing to be
optimized. For each choice of corner frequency, an LQG design would be
conducted and the settling time or some other measure would be used in the
cost function.
Any thoughts?
Thanks,
Ryan
--
Ryan Krauss, Ph.D.
Associate Professor
Mechanical Engineering
Southern Illinois University Edwardsville
|
