Discussion On: “decentralized Supervisory Control of Fuzzy Discrete Event Systems”

Yongzhi Cao

DOI: https://doi.org/10.3166/ejc.14.244-246

IF: 2.649

2008-01-01

European Journal of Control

Abstract:The paper by Liu and Qiu is concerned with thedecentralized supervisory control problem for fuzzydiscrete event systems (FDESs) with partial and fuzzyobservations. On the one hand, this work is a naturalcontinuation of the authors’ paper [8] in which thecentralized control under fuzzy observations was dis-cussed; on the other hand, it is a good extension andcomplement to some related results in [1]. In terms ofthe decentralized supervisory control, the key point ofdivergencebetweenthepaperand[1]isthattheformerallows observations to be fuzzy. The fuzziness ofobservations, originally suggested by Lin and Ying in[4,5], seems an interesting feature of FDESs anddesires formal study.In addition to introducing several necessary defi-nitions, the main technical contributions of the paperare a decentralized supervisory control theorem and acomputing approach to verifying the fuzzy co-observability condition in this theorem. The theoremshows that there exist non-blocking local fuzzysupervisors if and only if the fuzzy language to besynthesized is fuzzy controllable, fuzzy co-observable,and closed. Determining the existence of local fuzzysupervisors appeals to methods for checking the fuzzycontrollability and the fuzzy co-observability. Thepaper develops a feasible way to test the fuzzy co-observability condition, since another method forchecking the fuzzy controllability condition hasalready been introduced in [7]. Such a way is highlydependentonthemax-minfuzzyautomatonmodelingof FDESs. More explicitly, it is the max-min operatorin the model that makes the number of all statesreachable from any state in a system finite, whichclearly facilitates the use of computing tree.