Reconfigurable GSPNs: A modeling formalism of evolvable discrete-event systems

摘要

Nowadays, a wide range of systems are becoming structurally dynamic, variably interconnected, and highly complex. The use of classical formal approaches, such as Petri nets, in the design of such systems becomes neither convenient nor sufficient. Indeed, they cannot handle, in a natural way, the dynamic structure and the growing complexity of modern systems.Introducing reconfigurability in Petri nets, as well as generalized stochastic Petri nets increases the modeling power, but decreases the applicability of analysis techniques. In fact, several properties become undecidableIn this paper, we extend generalized stochastic Petri nets to a reconfigurable formalism, while maintaining verifiability with reduced complexity. The proposed approach identifies a set of properties that are preserved after reconfiguration. These properties are decidable with reduced time and space complexity. The use of the proposed formalism is illustrated through a running example. (C) 2019 Elsevier B.V. All rights reserved.