DocumentCode
486497
Title
Deadtime Approximations with Adaptive Deadtime Compensation
Author
Durbin, L.D.
Author_Institution
Chemical Engineering Department, Texas A & M University, College Station, Texas
fYear
1985
fDate
19-21 June 1985
Firstpage
1707
Lastpage
1712
Abstract
Simple polynomial approximations of the Padé type are considered for deadtime estimation with single-loop adaptive control. An LQG regulator is used with combined parameter identification via a Sequential Gauss-Newton (SGN) method. These include provisions for unknown constant load inputs and random noise effects. An equation error approach is used based upon the deadtime-plus-first order (DT+FOE) model form. However, for the error partial with respect to deadtime, two forms are given based upon a simple Padé approximation for part of the deadtime. One form of the partial relates to the equation error and the other is a hybrid form relative to an output+equation error formulation. Simulation tests of the adaptive control procedures were made for combined parameter changes, including the deadtime. These results are discussed relative to conditions for satisfactory convergence.
Keywords
Adaptive control; Convergence; Equations; Least squares methods; Newton method; Parameter estimation; Polynomials; Recursive estimation; Regulators; Testing;
fLanguage
English
Publisher
ieee
Conference_Titel
American Control Conference, 1985
Conference_Location
Boston, MA, USA
Type
conf
Filename
4788889
Link To Document