测控集群系统随机Petri网模型与可用性分析

研究并实现了一个应用于测控领域的高可用性实时集群系统，该系统充分采用冗余技术以消除系统中的单一故障点．首先阐述了系统的体系结构，然后在介绍随机Petri网基本概念的基础上，建立了系统的随机Petri网模型，并在此模型的基础上，分析了系统各部分的稳态可用度和系统的总稳态可用度．经理论计算表明，测控集群系统一年的平均故障停机时间约为3．44分钟，因此，系统具有较高的可用性，完全满足关键应用场所的高可用性要求．

Stochastic Petri Net Modeling and Availability Analysing of Observation and Control Cluster

Cluster is an important approach to get high availability, high scalability and performance cost rate in critical application field. This paper designs and implements a high availability real-time cluster system used in observation and control domain. All parts of the cluster are sufficiently redundant to avoid sigle failure point. First, the system architecture is explained. Then, the basic consepts of the stochastic Petri net (SPN) are introduced and the SPN of the system is modeled to formally describe all the parts. Finally, the steady-state availability of the parts and the total steady-state availability of the system are analysed. The result of the theory analyse proves that the average failure time in a year of the cluster is about 3.44 minutes. It shows that the cluster has fully high availability to be used in critical application field.