Information Rich Enough Language for Mealy Automata

碩士 === 國立臺灣科技大學 === 工程技術研究所自動化及控制學程 === 90 === The dynamic behaviors of an input/output discrete event system are: the states of the system will change according to its input events and send output events as the response. Generally, in framework of the supervisory system with exchanged 「...

Full description

Bibliographic Details
Main Authors: Shi-Tong Wang, 汪世通
Other Authors: Sheng-Luen Chung
Format: Others
Language:en_US
Published: 2002
Online Access:http://ndltd.ncl.edu.tw/handle/74453679828165474594
Description
Summary:碩士 === 國立臺灣科技大學 === 工程技術研究所自動化及控制學程 === 90 === The dynamic behaviors of an input/output discrete event system are: the states of the system will change according to its input events and send output events as the response. Generally, in framework of the supervisory system with exchanged 「command」 and 「response」 event, the plant looks command as input, and looks response as output. On the other hand, supervisor looks response as input, and looks command as output. The objective of discrete event system identification problems is: given an unknown discrete event system, we want to reconstruct the system from its input/output of finite length. Thus, this research can be classified into three layers: (1) given the finite input/output strings of an unknown system, we want to construct an input/output state machine that has the same input/output strings. In addition to describe above, we hope the constructed automaton is “minimal”. (2) We hope that can conjecture the dynamic behaviors for the given unknown system from the finite-length testing strings. In the concepts of formal languages, we want to figure out the system languages from the strings of the constructed automaton. Due to this, the objective is to find an automaton and the testing strings that can help us to reconstruct the language of the given unknown system. Moreover, the found testing string is “minimal”. (3) For realization, we want to infer an online algorithm that can correct our estimated state machine incrementally form the new continuously observed data so that the behaviors of the constructed state machine will much conform with the given system. Furthermore, if the observed string satisfies with the sufficient condition of (2), then the language of the reconstructed system is exactly the same as the unknown given system.