基于正态Petri网的模型参数识别研究  

Research on Model Parameter Identification Based on Normal Stochastic Petri Net

作　　者：陈哲[1] 刘久富[1] 王正谦[1] 

机构地区：[1]南京航空航天大学自动化学院,江苏南京210016

出　　处：《计算机技术与发展》2014年第5期20-22,共3页Computer Technology and Development

基　　金：国家自然科学基金资助项目(60674100);南京航空航天大学青年科学创新基金(NS2010069)

摘　　要：时间因素在对实时系统的分析中占据着重要的位置。随着时间因素在Petri网应用分析中的地位越发重要,各种与时间相关的Petri网模型相继被提出。文中结合动态实时系统的运行特征,设计了一种变迁点火时间服从正态分布的随机Petri网模型,并基于对系统行为的统计分析,提出了一种用于对变迁时间参数进行识别的数值算法。以一种随机制造单元为例,建立正态随机Petri网模型,并运用参数识别算法确定变迁时间参数。仿真结果符合系统的实际运行结果,证明了该方法的有效性。The factor of time had played an important role in the analysis of real-time systems. As the importance of time analysis in the application of Petri nets is increasing,a variety of timed Petri net models were proposed. Considering the characteristics of dynamic real-time systems,a stochastic Petri net with normal distributions of the firing periods was designed and a method for the identification of the parameters was proposed based on a statistical analysis of the behavior of the system. The system model of a random manufacturing unit was built with normal Petri net and the parameters of firing periods were verified by the estimation algorithm. The simulation results fit in-to the actual behavior of the system and prove the effectiveness of this method.

关 键 词：非马尔可夫网 随机PETRI网 模型设计 参数识别 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]