| Summary: | 碩士 === 國立中興大學 === 土木工程學系 === 84 === ABSTRACTLinear quadratic
regulator has been used extensively in many control
systemsdesigned for structural control applications due to its
stability androbustness.However, the results obtained from
simulations, model experimentsand full-scale structural
applications show that it is difficult to employquadratic
performance criteria and linear feedback control lawsto produce
a significant peak response reduction when the peak response
occursduring the first few cycles of the time history,which is
usually the caseunder seismic ground excitations. Since peak
response is closely related tostructural safety,control
algorithms which provide improved peak responsereduction are
desirable. A class of nonlinear control law is presentedin this
repor for this purpose. It is showen that nonlinear control
lawcan significantly improve peak response reduction under the
same constraintsimposed on the control resources as in the
linear quadratic regulator case.In real active control systems,
time is consumed in the acquisition of response and excitation
data,data processing, on-line calculation, and control force
execution.Most of previous studies neglected the time delay
effect based on theaugument that flexible structures usually
have a fairly long naturalperiod compared with the delay time,
which makes the time delay effectnegligible. However,The delay
time may be minimized by employing more advanced hardware
andsoftware. But, time delay cannot be avoided and eliminated
even withpresent-day technology. Small delay timenot only can
render the control ineffective, but also may cause the
systeminstability. Hence, time delay effect must be considered
in control designbefore real implementation of active control
can be put into action.For this purpose, in this study, the
full-order mathematical model ofa structure is considered to
develop the optimal time-delayed directoutput feedback control
algorithm for discrete-time structural systems.Optimal output
feedback gains are obtained through variational processsuch that
certain prescribed quadratic performance index is minimized.They
are calculated by solving three linear algebraic
equationssimultaneously using Kronecker algebra. And then,
discrete-time control forcesare obtained from the multiplication
of output measurements by thesepre-calculated feedback gains.
|
|---|
