The behavior of real-time systems must meet the strict constraints of time and consistent use of the data to ensure its reliability. We propose in this work the filter patterns for a consistent modeling of the data used by real-time system components. These patterns must ensure the temporal validity and strong cohesion of the data used by the system components. We associate to the UML activity diagram a graph of data of the system that uses a data filter. A filter of data accepts or rejects the data used by a component for keeping its computing consistent. We formalize our pattern filter by translating them in Petri networks. We validate our results by applying them to a regulator simplified speed that we simulate in CPN tools.

formal methods, Petri networks, real-time systems, consistent system, pattern.