Engineering Mechanics

Proceedings Vol. 10 (2004)

ENGINEERING MECHANICS 2004

NATIONAL CONFERENCE WITH INTERNATIONAL PARTICIPATION
May 10 – 13, 2004, Svratka, Czech Republic

;

Editors: Igor Zolotarev and Alena Poživilová

ISBN 80-85918-88-9 (printed)
ISSN 1805-8248 (printed)
ISSN 1805-8256 (electronic)

list of papers scientific commitee

Use of continuous action reinforcement learning automata for asynchronous electromotor control

Březina T., Turek M.
pages 53 - +9p., full text

Relatively unknown reinforcement learning algorithm, so called continuous action reinforcement learning automaton, is presented in this contribution. Automaton learning algorithm is based on rewarding, that gradually evolves set of probability densities. This set is consequently used for action set determination. Simulation study describing learning and behavior of asynchronous electromotor control is further presented. Standard PSD controller is used whose parameter values represent actions of three independent automata. The aim of online learning process is to minimize mean square of control error. Here described learning algorithm is simple to implement, robust to high level of noise.

back to list of papers

Text and facts may be copied and used freely, but credit should be given to these Proceedings.

All papers were reviewed by members of the scientific committee.

Engineering Mechanics

International Conference

Proceedings Vol. 10 (2004)

ENGINEERING MECHANICS 2004