Summary
International Symposium on Nonlinear Theory and its Applications
2008
Session Number:C4L-B
Session:
Number:C4L-B2
Reinforcement Learning of Optimal Supervisor for Timed Discrete Event Systems
Tatsushi Yamasaki, Toshimitsu Ushio,
pp.-
Publication Date:2008/9/7
Online ISSN:2188-5079
Summary:
Timed discrete event systems are a class of discrete event systems which can represent information of time of event occurrences. In this paper, we propose a synthesis method of a supervisor for such systems. The supervisor is constructed by reinforcement learning under the framework of the supervisory control proposed by Brandin and Wonham, and is optimal with respect to a language measure which is introduced by Wang and Ray. The proposed method is applicable to the situation which precise information of the system is unknown.