LL_GF_MATHMOD_feb_2009

Liu, L.; Frey, G.:
Efficient Simulation of Hybrid Control Systems in MODELICA/DYMOLA,
Proceedings of the 6th Vienna International Conference on Mathematical Modelling (MATHMOD 2009), Vienna, Austria, pp. 1344-1352, February 11 - 13, 2009.

Abstract:
Though systems with mixed discrete/continuous behaviors can be handled by proper hybrid simulation tools, the efficiency of simulation is often not satisfactory. The performance of simulation is affected by various system properties such as the frequency of events, the number of continuous state variables, etc. In an ordinary modeling and simulation approach, the hybrid sys-tem is considered as a whole thus every local event has global effects. I.e. An event triggered by any element of the system causes the numerical solver to detect the event and recalculate all state variables. This works well if the discrete event logic couples tightly with continuous behavior evo-lution. A tight coupling means that the discrete event sub-system governs the whole system and all its events necessarily trigger the switch of equation system or stop/restart of the numerical solver. However, in case of a loose coupling, i.e. not all the events from the discrete event sub-system are relevant to the continuous sub-system, this approach is excessively expensive. This paper gives a general approach to achieve more efficient simulations of hybrid systems with loose coupling, es-pecially for stiff systems such as the Networked Control/Automation Systems (NCS/NAS) in which different system components have strongly distinguished temporal characteristics. Advan-tages of this approach are demonstrated using the simulation tool Modelica/Dymola.

Keywords: Modelica, Hybrid Control System, Efficient Simulation