Advanced

Dual Control of an Integrator Using Neuro-Dynamic Programming

Cervin, Anton LU (1998) In MSc Theses
Department of Automatic Control
Abstract
The dual control problem for an integrator with time-varying gain is reviewed. The neuro-dynamic programming technique of approximate policy iteration is explained. It is shown how a version of this technique, which employs a Gaussian radial basis function network, can be used to calculate a good sub-optimal dual controller for the integrator. This controller performs much better than the certainty equivalence and the cautious controllers. Finally, extensions to more realistic problems are discussed.
Please use this url to cite or link to this publication:
author
Cervin, Anton LU
supervisor
organization
year
type
H3 - Professional qualifications (4 Years - )
subject
publication/series
MSc Theses
report number
TFRT-5598
ISSN
0280-5316
language
English
id
8848550
date added to LUP
2016-03-24 17:34:39
date last changed
2016-03-24 17:34:39
@misc{8848550,
  abstract     = {The dual control problem for an integrator with time-varying gain is reviewed. The neuro-dynamic programming technique of approximate policy iteration is explained. It is shown how a version of this technique, which employs a Gaussian radial basis function network, can be used to calculate a good sub-optimal dual controller for the integrator. This controller performs much better than the certainty equivalence and the cautious controllers. Finally, extensions to more realistic problems are discussed.},
  author       = {Cervin, Anton},
  issn         = {0280-5316},
  language     = {eng},
  note         = {Student Paper},
  series       = {MSc Theses},
  title        = {Dual Control of an Integrator Using Neuro-Dynamic Programming},
  year         = {1998},
}