含有非马尔可夫过程的排队Petri网模型和性能分析

A Model of Queuing Petri Net Including Non-Markovian Process and Performance Analysis

本文提出了一种新的高级性能模型技术 ,称作NM QPN(含有非马尔可夫过程的排队Petri网 ,QueuingPetriNetincludingNon Markovianprocesses) ,它综合了排队网 ,随机Petri网以及模拟求解各自的特点 .NM QPN以模拟模型为总体框架 ,发挥排队网和随机Petri网各自的优势对系统进行建模 .提出了一套完整的NM QPN模型求解方案 ,通过流等价方法将模型中的马尔可夫过程进行化简 ,这样可以大大减少模型的状态 。

The technique of Stochastic Petri Nets(SPN) was wildly used in the field of performance evaluation in the past 20 years.But the SPN model is still in face of the problem of explosion of the state space.And if in stochastic modeling the idealized assumption of exponential distribution is removed,the resulting stochastic process is non Markovian.In this paper,we present Queuing Petri Net including Non Markovian processes (NM QPN),a novel high level modeling technique,which is based on the former research.This modeling technique integrates the characteristics of Queuing Nets,Stochastic Petri Nets and simulation solving.A complete solving scheme of NM QPN was presented.The algorithm of Flow Equivalent is used to analyze and simplify the Markovian processes parts in the model so that the state space will be greatly reduced,and then simulation technique to solve the remaining non Markovian processes was used.