利用基本信标的时延Petri网的死锁预防策略

A deadlock prevention policy for a class of timed Petri net using elementary siphons

针对一类时延Petri网的死锁预防问题,提出了一个有效的控制算法.将基于基本信标的死锁预防方法应用到时延Petri网中,不需要列举全部可达状态,通过只对基本信标添加控制库所和根据需要调整控制深度变量实现死锁预防,而且最后的控制器为活的网系统.首先将时延Petri网拉伸,使得不改变原来系统性能的情况下,拉伸后的网SPN所有变迁均为单位时延,然后针对SPN,采用基本信标控制方法,根据所得标识约束添加控制库所,使基本信标和从属信标均为P-不变式可控,最后还原时延Petri网,得到活的、结构简单的Petri网监督控制器.

To solve the problem of deadlock prevention for timed Petri nets, an effective prevention policy based on elementary siphons is proposed in this paper. Without enumerating all the reachable markings, deadlock prevention is achieved by adding monitors for each elementary siphon and increasing control depth variants where needed. And the final supervisor is live. First, a timed Petri net is stretched into a SPN. Unchanging the system performance, each transition in the SPN has an unit time delay. Then ignoring the factor of delay time, the elementary siphon-control-based approach is applied. Monitors based on the marking constraints are added to the plant model to ensure all elementary and dependent siphons invariant-controlled. A liveness-enforcing supervisor with simple structure can be realized by reverting the TdPN.