Neuro-fuzzy learning of strategies for optimal control problems

Kaivan Kamali, Lijun Jiang, John Yen, K. W. Wang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Various techniques have been proposed to automate the weight selection process in optimal control problems; yet these techniques do not provide symbolic rules that can be reused. We propose a layered approach for weight selection process in which Q-learning is used for selecting weighting matrices and hybrid genetic algorithm is used for selecting optimal design variables. Our approach can solve problems that genetic algorithm alone cannot solve. More importantly, the Q-learning's optimal policy enables the training of neuro-fuzzy systems which yields reusable knowledge in the form of fuzzy if-then rules. Experimental results show that the proposed method can automate the weight selection process and the fuzzy if-then rules acquired by training a neuro-fuzzy system can solve similar weight selection problems.

Original languageEnglish (US)
Title of host publicationNAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society
Pages199-204
Number of pages6
DOIs
StatePublished - Dec 1 2005
EventNAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society - Detroit, MI, United States
Duration: Jun 26 2005Jun 28 2005

Publication series

NameAnnual Conference of the North American Fuzzy Information Processing Society - NAFIPS
Volume2005

Other

OtherNAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society
CountryUnited States
CityDetroit, MI
Period6/26/056/28/05

Fingerprint

Neuro-fuzzy
Fuzzy systems
Optimal Control Problem
Genetic algorithms
Fuzzy If-then Rules
Neuro-fuzzy Systems
Q-learning
Hybrid Genetic Algorithm
Optimal Policy
Weighting
Genetic Algorithm
Strategy
Learning
Experimental Results
Training
Optimal design

All Science Journal Classification (ASJC) codes

  • Computer Science(all)
  • Mathematics(all)

Cite this

Kamali, K., Jiang, L., Yen, J., & Wang, K. W. (2005). Neuro-fuzzy learning of strategies for optimal control problems. In NAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society (pp. 199-204). [1548533] (Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS; Vol. 2005). https://doi.org/10.1109/NAFIPS.2005.1548533
Kamali, Kaivan ; Jiang, Lijun ; Yen, John ; Wang, K. W. / Neuro-fuzzy learning of strategies for optimal control problems. NAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society. 2005. pp. 199-204 (Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS).
@inproceedings{ecf70d4ae23142d9a6a2588ebba0f824,
title = "Neuro-fuzzy learning of strategies for optimal control problems",
abstract = "Various techniques have been proposed to automate the weight selection process in optimal control problems; yet these techniques do not provide symbolic rules that can be reused. We propose a layered approach for weight selection process in which Q-learning is used for selecting weighting matrices and hybrid genetic algorithm is used for selecting optimal design variables. Our approach can solve problems that genetic algorithm alone cannot solve. More importantly, the Q-learning's optimal policy enables the training of neuro-fuzzy systems which yields reusable knowledge in the form of fuzzy if-then rules. Experimental results show that the proposed method can automate the weight selection process and the fuzzy if-then rules acquired by training a neuro-fuzzy system can solve similar weight selection problems.",
author = "Kaivan Kamali and Lijun Jiang and John Yen and Wang, {K. W.}",
year = "2005",
month = "12",
day = "1",
doi = "10.1109/NAFIPS.2005.1548533",
language = "English (US)",
isbn = "078039187X",
series = "Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS",
pages = "199--204",
booktitle = "NAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society",

}

Kamali, K, Jiang, L, Yen, J & Wang, KW 2005, Neuro-fuzzy learning of strategies for optimal control problems. in NAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society., 1548533, Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS, vol. 2005, pp. 199-204, NAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society, Detroit, MI, United States, 6/26/05. https://doi.org/10.1109/NAFIPS.2005.1548533

Neuro-fuzzy learning of strategies for optimal control problems. / Kamali, Kaivan; Jiang, Lijun; Yen, John; Wang, K. W.

NAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society. 2005. p. 199-204 1548533 (Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS; Vol. 2005).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Neuro-fuzzy learning of strategies for optimal control problems

AU - Kamali, Kaivan

AU - Jiang, Lijun

AU - Yen, John

AU - Wang, K. W.

PY - 2005/12/1

Y1 - 2005/12/1

N2 - Various techniques have been proposed to automate the weight selection process in optimal control problems; yet these techniques do not provide symbolic rules that can be reused. We propose a layered approach for weight selection process in which Q-learning is used for selecting weighting matrices and hybrid genetic algorithm is used for selecting optimal design variables. Our approach can solve problems that genetic algorithm alone cannot solve. More importantly, the Q-learning's optimal policy enables the training of neuro-fuzzy systems which yields reusable knowledge in the form of fuzzy if-then rules. Experimental results show that the proposed method can automate the weight selection process and the fuzzy if-then rules acquired by training a neuro-fuzzy system can solve similar weight selection problems.

AB - Various techniques have been proposed to automate the weight selection process in optimal control problems; yet these techniques do not provide symbolic rules that can be reused. We propose a layered approach for weight selection process in which Q-learning is used for selecting weighting matrices and hybrid genetic algorithm is used for selecting optimal design variables. Our approach can solve problems that genetic algorithm alone cannot solve. More importantly, the Q-learning's optimal policy enables the training of neuro-fuzzy systems which yields reusable knowledge in the form of fuzzy if-then rules. Experimental results show that the proposed method can automate the weight selection process and the fuzzy if-then rules acquired by training a neuro-fuzzy system can solve similar weight selection problems.

UR - http://www.scopus.com/inward/record.url?scp=33745000736&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33745000736&partnerID=8YFLogxK

U2 - 10.1109/NAFIPS.2005.1548533

DO - 10.1109/NAFIPS.2005.1548533

M3 - Conference contribution

AN - SCOPUS:33745000736

SN - 078039187X

SN - 9780780391871

T3 - Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS

SP - 199

EP - 204

BT - NAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society

ER -

Kamali K, Jiang L, Yen J, Wang KW. Neuro-fuzzy learning of strategies for optimal control problems. In NAFIPS 2005 - 2005 Annual Meeting of the North American Fuzzy Information Processing Society. 2005. p. 199-204. 1548533. (Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS). https://doi.org/10.1109/NAFIPS.2005.1548533