Publication detail

Q-LEARNING USED FOR CONTROL OF AMB: REDUCED STATE DEFINITION

BŘEZINA, T., KREJSA, J.

Original Title

Q-LEARNING USED FOR CONTROL OF AMB: REDUCED STATE DEFINITION

Type

conference paper

Language

English

Original Abstract

A great intention is lately focused on Reinforcement Learning (RL) methods. Previous work showed that stochastic strategy improved model free RL method known as Q-learning used on active magnetic bearing (AMB) model. So far the position, velocity and acceleration were used to describe the state of the system. This paper shows simplified version of controller which uses reduced state definition - position and velocity only. Furthermore the controlled initial conditions area and its development during learning are shown. Numerical experiments proved that simplified controller version is fully capable of AMB control.

Keywords

Reinforcement Learning, Q-learning, Active Magnetic Bearing

Authors

BŘEZINA, T., KREJSA, J.

RIV year

2002

Released

5. 6. 2002

Publisher

Brno University of Technology, Faculty of Mechanical Engineering

Location

Brno

ISBN

80-214-2135-5

Book

Mendel 2002

Edition number

1

Pages from

347

Pages to

352

Pages count

6

BibTex

@inproceedings{BUT10054,
  author="Tomáš {Březina} and Jiří {Krejsa}",
  title="Q-LEARNING USED FOR CONTROL OF AMB: REDUCED STATE DEFINITION",
  booktitle="Mendel 2002",
  year="2002",
  number="1",
  pages="6",
  publisher="Brno University of Technology, Faculty of Mechanical Engineering",
  address="Brno",
  isbn="80-214-2135-5"
}