Skip to Main content Skip to Navigation
Theses

Model-driven co-simulation of Cyber-Physical Systems

Abstract : Cyber Physical Systems (CPS) are integrations of physical and computational components. CPS are difficult to model and verify because the heterogeneous nature of their components requires many different modeling formalisms. The global verification of the system can be achieved by co-simulation. FMI standard offers a standard interface to couple two or more simulators in a co-simulation environment, known as master. This latter is responsible for providing an algorithm with efficient orchestration and synchronization of the involved components, known as FMUs. However, FMI was originally intended for co-simulation of physical processes, with limited support for formalisms such as DE and Dara-Flow, even if this kind of formalisms are commonly used to model the logic of software parts of a system. In particular, while UML is the reference standard for software modeling and is very commonly used in industry, none of the present-day FMI-based co-simulation solutions consider UML models. Our thesis is that system engineering in general would greatly benefit from the consideration of UML in FMI-based co-simulation approach. It would indeed enable a significant number of software designers to evaluate the behavior of their software components in their simulated environment, as soon as possible in their development processes, and therefore make early and better design decisions. It would also open new interesting perspectives for CPS system engineers, by allowing them to consider a widely used modeling language for the software parts of their systems. In this context, the objective of this work is to define an FMI-based co-simulation environment for CPS with integration of UML models for software part. Our contribution is twofold: locally at the level of UML models, and globally at the master level. At the local level, we set up an incremental approach where we address different kinds of discrete event systems characterizing the computational components. We base our proposals on OMG standards fUML and PSCS which define precise execution semantics for a subset of UML. They provide an interesting and formal basis for the integration of UML models in CPSs co-simulation approaches. For each kind of system, we first identify a set of rules to model it with UML and potential extensions to fUML in case where execution semantics of the required UML elements are not defined by fUML. Then, at the global level, we propose a master algorithm for each kind of systems. The proposed masters take into account not only external and internal dependencies between components and their capabilities, but also and especially their models of time. They rely on adaptation of fUML semantics to that of the FMI API. Based on these adaptations, the master algorithm is able both to propagate data between components and to trigger them at the correct points of time. The approach is illustrated with use cases from the energy domain where the purpose is to verify energy management strategies defined as software components at different levels of the control module of an energy system.
Document type :
Theses
Complete list of metadatas

Cited literature [56 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-01865810
Contributor : Abes Star :  Contact
Submitted on : Sunday, September 2, 2018 - 1:05:33 AM
Last modification on : Wednesday, October 14, 2020 - 3:59:20 AM
Long-term archiving on: : Tuesday, December 4, 2018 - 10:16:21 AM

File

76297_GUERMAZI_2017_archivage....
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-01865810, version 1

Collections

Citation

Sahar Guermazi. Model-driven co-simulation of Cyber-Physical Systems. Modeling and Simulation. Université Paris Saclay (COmUE), 2017. English. ⟨NNT : 2017SACLS333⟩. ⟨tel-01865810⟩

Share

Metrics

Record views

299

Files downloads

640