Semantics of multi-mode DAE systems - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Other Publications Year : 2013

Semantics of multi-mode DAE systems


Hybrid systems modelers exhibit a number of difficulties related to the mix of continuous and discrete dynamics and sensitivity to the discretization scheme. Modular modeling, where subsystems models can be simply assembled with no rework, calls for using Differential Algebraic Equations (DAE). In turn, DAE are strictly more difficult than ODE. They require sophisticated pre-processing using various notions of index before they can be submitted to a solver. In this report we study some fundamental issues raised by the modeling and simulation of hybrid systems involving DAEs. The objective of this work is to serve for the evolution and the design of future releases of the Modelica language for such systems. We focus on the following questions: * What is the proper notion of index for a hybrid DAE system? * What are the primitive statements needed for a DAE hybrid systems modeler? The differentiation index for DAE explicitly relies on everything being differentiable. Therefore, generalizations to hybrid systems must be done with caution. We propose relying on non-standard analysis for this. Non-standard analysis formalizes differential equations as discrete step transition systems with infinitesimal time basis. We can thus bring hybrid DAE systems to their nonstandard form, where the notion of difference index can be firmly used. From this study, general hints for future releases of Modelica can be drawn.


Other [cs.OH]
Fichier principal
Vignette du fichier
MODRIO_PublicReport_D411.pdf (803.88 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-00938891 , version 1 (29-01-2014)


  • HAL Id : hal-00938891 , version 1


Albert Benveniste, Timothy Bourke, Benoît Caillaud, Marc Pouzet. Semantics of multi-mode DAE systems. 2013. ⟨hal-00938891⟩
416 View
152 Download


Gmail Mastodon Facebook X LinkedIn More