Automatic core design using reinforcement learning

Yoko Kobayashi, Eitaro Aiyoshi

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

2 Citations (Scopus)


This paper deals with the application of multi-agents algorithm to the core design tool in a nuclear industry. We develop an original solution algorithm for the automatic core design of boiling water reactor using multi-agents and reinforcement learning. The characteristics of this algorithm are that the coupling structure and the coupling operation suitable for the assigned problem are assumed, and an optimal solution is obtained by mutual interference in multi state transitions using multi-agents. We have already proposed an integrated optimization algorithm using a two-stage genetic algorithm for the automatic core design. The objective of this approach is to improve the convergence performance of the optimization in the automatic core design. We compared the results of the proposed technique using multi-agents algorithm with the two-stage genetic algorithm that had been proposed before. The proposed technique is shown to be effective in reducing the iteration numbers in the search process.

Original languageEnglish
Title of host publicationProceedings of the 2004 American Control Conference (AAC)
Number of pages6
Publication statusPublished - 2004 Nov 29
EventProceedings of the 2004 American Control Conference (AAC) - Boston, MA, United States
Duration: 2004 Jun 302004 Jul 2

Publication series

NameProceedings of the American Control Conference
ISSN (Print)0743-1619


OtherProceedings of the 2004 American Control Conference (AAC)
Country/TerritoryUnited States
CityBoston, MA

ASJC Scopus subject areas

  • Electrical and Electronic Engineering


Dive into the research topics of 'Automatic core design using reinforcement learning'. Together they form a unique fingerprint.

Cite this