基于一类特殊s3pr网的多项式复杂度死锁避免策略分析-deadlock avoidance strategy analysis of polynomial complexity based on a special s3pr network.docxVIP

摘要本文致力于研究 S3PR 网的死锁控制问题，得到具有最大行为许可性和多项 式复杂度的控制策略是死锁研究的主要目的。本文提出为每个严格极小信标添加一个控制深度为一的控制库所的死锁预防 策略。基于可达图，分析该策略下一个网存在活性控制器的充要条件是该网的可 达图中不存在不可控标识，这类网称为可行网。并且为可行网添加控制库所得到 的活性控制器具有最大行为许可性。不可控标识判断的主要理论依据是，在 S3PR 网的信标清空前状态下，一定可以内部发射。为了方便研究，本文首次提出信标 的清空状态、清空前状态、核心资源及相邻资源等概念。一般情况下，即使是最简单的自动制造系统，计算其最大行为许可的死锁避 免策略都是 NP-hard 的，有文献提出一种判断资源变迁回路饱和的死锁避免策略， 并证明了对于不含有单位容量的中心资源的 S3PR 网，该策略具有多项式复杂度。 本文通过研究资源变迁回路和信标的关系，说明了该策略同样可以应用于可行网； 本文提出弱从属 S3PR 网的概念，从结构上分析了该类网具有的特殊性质，并证 明该策略可以应用于某些特殊的弱从属 S3PR，且具有多项式复杂度。关键词：弱从属 S3PR 网资源变迁回路不可控标识最大行为许可 多项式复杂度AbstractThis thesis focuses on deadlock control for S3PR nets. The main purpose of deadlock control is to find a maximally permissive control policy that is of polynomial complexity.A deadlock prevention policy that adds a new place whose control depth is one for every strict minimal siphon is proposed. The necessary and sufficient condition of deriving a liveness-enforcing supervisor for an S3PR net under this policy is obtained by analyzing its reachability graph. The condition is that the reachability graph of the S3PR has no obstinate markings that are determined on the basis of the fact that an internal transition of a siphon must be enabled in its empty-before marking. A net which has a liveness-enforcing supervisor is called a feasible net and its supervisor is maximally permissive. The concepts of empty-before markings, empty markings, key resources and neighboring resources are proposed for the first time in this paper.Even for a simple automated manufacturing system, the computation of a maximally permissive deadlock avoidance policy is in sense NP-hard. A maximally permissive deadlock avoidance policy for an S3PR net without a resource that has unit capacity is proposed. This policy is implemented by means of deciding whether a saturated recourse-transition circuit exists in a subsequent marking. Based on the relationship between the resource-transition circuits and the siphons, it is proved that this policy can be applied to feasible ne