分布式离散事件系统的模式故障预测研究

针对分布式离散事件系统, 提出了一种模式故障预测方法。首先对分布式离散事件系统的模式故障可预测性进行形式化。通过构造一个模式故障识别器, 从系统所有行为中识别出所发生的模式故障, 并针对分布式系统的不同观测点构造不可观测闭包。在此基础上, 联合各站点观测到的事件序列构造出模式故障预测验证器, 解决了分布式离散事件系统的模式故障预测问题。得出了一个判定分布式离散事件系统模式故障可预测性的充分必要条件, 并提出了相应的模式故障预测算法, 实现了对分布式离散事件系统的模式故障预测。最后, 对分布式模式故障预测验证器的构建和模式故障预测算法进行了复杂度分析。

A Research on Patterns Fault Prediction of Decentralized Discrete Event Systems

In recent years, the research on fault prediction of discrete event systems has received considerable attention. In this research, the prediction of patterns fault for decentralized discrete event systems is investigated. Firstly, the notion of predictability of patterns fault for decentralized systems is defined and the patterns fault predictability of distributed discrete event systems formalized. By constructing a patterns fault recognizer, patterns fault is identified from all behaviors of the system, and the unobservable closure is constructed for different observation points of decentralized system. On this basis, the patterns fault prediction verifier is constructed by combining the events observed at each station, which solves the problem of patterns fault prediction of decentralized discrete event system. A sufficient and necessary condition for determining the predictability of the decentralized systems is obtained. The corresponding patterns fault prediction algorithm is proposed and implemented. Finally, the complexity analyses of the construction of the decentralized patterns fault prediction verifier and the patterns fault prediction algorithm are carried out.