Research On Simulation And Output Analysis Of Discrete Event Systems With Fuzzy Parameters

Discrete event system modeling and simulation has been increasingly used in the system analysis and decision support during the last few decades. Classical discrete event system simulation methodology usually assumes that sufficient information (e.g. parameters of input processes, probability distributions of random variables in the system, etc.) is available to model the real system precisely. However, this is a stringent requirement in most practice. Therefore, the problem to address imprecise information during discrete event system simulation is worth studying.Fuzzy set and possibility theory has been proposed for about forty years, and has been regarded as a powerful tool for representing and dealing with imprecision. Thus, the study of the methodology for fuzzy discrete event system simulation draws considerable interest of simulation community. However, there exists great difference between fuzzy discrete event system simulation and its classical counterpart. This paper focuses on a type of fuzzy discrete event system simulation with fuzzy input parameters, and a new methodology is proposed based on the recently developed random fuzzy theory. This mainly includes three parts: 1) the method for simulation, 2) the method for output data analysis, and 3) the method for influence evaluation of input fuzziness on output performance measures.The first part deals with the simulation execution of such systems, with the aim at estimating the membership function of the output performance measures. This paper proposed a method for fuzzy discrete event system simulation which avoids the problem of time paradox. Instead of obtaining a crisp value in classical discrete event simulation, fuzzy simulation may yield a fuzzy number to represent the output measure, and this method will estimate the membership function of the output performance effectively. The calculation of output measures is basic of the subsequent analysis process. The second part arrives at analysis methods after simulation runs. Output analysis is an important element in system study through discrete event system simulation. In classical discrete event system simulation methodology, it involves: output data analysis for a single system, comparing alternative systems, and simulation-based optimization. This paper only discusses the output analysis for single systems in fuzzy discrete event system simulation. The third part comes into a new topic that only concerns fuzzy models, that is, how to evaluate the influence of input fuzziness on output measures. The method addressing this problem will assist in the process of system modeling and design.Every method is followed by simulation experiments to attest its effectiveness, and the results show that the new approach proposed in this paper yields more reasonable results. Finally, the effect of these new methods is exemplified by the simulation and analysis of a real production system.