Plenary lecture – Carla Seatzu
Modeling, analysis and intelligent control of automated manufacturing systems using Petri nets
Automated manufacturing systems (AMSs) may be defined as discrete production systems in which the handled materials are discrete entities, e.g. parts that are processed or assembled. As a result, AMSs can be effectively represented as discrete event systems (DESs) whose dynamics depends on the interaction of asynchronous discrete events, such as the arrival or departure of parts or products in a buffer, the start of an operation, the completion of a task and the failure of a machine. Petri nets (PNs) are one of the DES formalisms that most effectively allow to model AMSs, thanks to a series of key features: they are both a graphical and a mathematical formalism; they provide a compact and modular representation of large and complex systems; they allow to explicitly represent the notion of concurrency, namely activities that can be performed in parallel; the state is a vector that allows to solve a variety of problems using integer programming; finally, they allow to deal with systems having an infinite state space.
Several classes of PN models have been defined in the literature in the last decades, including Place/Transition nets, timed PNs, continuous PNs, hybrid PNs, colored PNs, fuzzy PNs, and so on. All such classes revealed particularly suited for the solution of relevant problems in AMSs, such as modeling, simulation, deadlock analysis, supervisory control, performance analysis and optimization, as well as problems related to the partial observability of the system state, in particular, fault diagnosis and diagnosability analysis, both in a centralized and in a decentralized framework.
The goal of this talk is twofold. First, a survey of the most significant contributions in this framework is provided. Then, future trends and open issues are discussed.