Presentation 2015-03-05
Optimal LLP supervisory control based on the learning of state transition model
Hijiri UMEMOTO, Tatsushi YAMASAKI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The authors have proposed an optimal LLP supervisory control method based on reinforcement learning for discrete event systems composed of subsystems. In this paper, we extend the previous work to estimate state transition probability model. It makes it possible to apply the method without cost information and state transition model a priori. In addition, we consider the case that the states of different subsystems change at the same time by the occurrence of an event. We also propose a method of dynamic adjustment of the number of lookahead steps in consideration of the real-time constraints of the system.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) supervisory control / discrete event system / limited lookahead policy / optimal control / reinforcement learning
Paper # MSS2014-92
Date of Issue

Conference Information
Committee MSS
Conference Date 2015/2/26(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Mathematical Systems Science and its applications(MSS)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Optimal LLP supervisory control based on the learning of state transition model
Sub Title (in English)
Keyword(1) supervisory control
Keyword(2) discrete event system
Keyword(3) limited lookahead policy
Keyword(4) optimal control
Keyword(5) reinforcement learning
1st Author's Name Hijiri UMEMOTO
1st Author's Affiliation Graduate School of Science and Technology, Setsunan University()
2nd Author's Name Tatsushi YAMASAKI
2nd Author's Affiliation Faculty of Science and Technology, Setsunan University
Date 2015-03-05
Paper # MSS2014-92
Volume (vol) vol.114
Number (no) 493
Page pp.pp.-
#Pages 6
Date of Issue