Contribution and challenges of multiagent simulation for factory digital twin

Abstract : A key technology for the Industry of the Future One of top 10 strategic technology trends for 2019 (cf. Gartner) A factory digital twin serves as a virtual replica of what is actually happening on the factory floor in near-real time SOLYSTIC SOSi TM SOLYSTIC is a provider of solutions for mail and parcel industry SOSi TM is an inhouse developped factory/supply chain digital twin; simulates 1 production year in 10 minutes Enables digital VSM and optimization of processes A solution for making easier the transition FROM Simulation as a tool for Observation: to understand the behavior of the reference system thanks to a model that is considered as a miniature reproduction of the reference system TO Simulation as a tool for Validation: to test an hypothesis of the reference system, to validate or to certify the underlying theory. Why? A CPS System is often Complex: the global behavior of the system is hard to model and any modifications is difficult Multiagent solution: The global system is not explicitly designed Multiagent solution: The multiagent concepts can be understood by non-experts Open: new component may be added to the system leading to a modification of the initial model Multiagent solution: only the new components and its interaction with other components must be designed Heterogeneous: different decision models, data models, … Multiagent solution: the resulting model is independent of the domain and can integrate several point of views. Decentralized: many local decisions without a centralized control Multiagent solution: decentralization management is the core of a multiagent system.