State Estimation of Partially Observable DES Using Fuzzy Timed Petri Nets

This paper addresses state estimation of partially observable discrete event systems exhibiting variations on the duration of activities; the system state is approximated by the marking of Fuzzy timed Petri Net (Ft P N) models. For this purpose a definition of Ft P N is presented in which fuzzy sets are associated to non-observable places for representing the ending time uncertainty of activities. The marking uncertainty for two subclasses of Ft P N is analyzed: marked graphs and state machines. Afterwards the marking uncertainty bound of such models including observable places is studied for obtaining an evaluation on the quality of the state estimation.