An Approach to Model-driven Architecture Applied to Hybrid Systems

Hybrid systems are characterized by a composition of discrete and continuous dynamics. In particular, the system has a continuous evolution and occasional jumps. The jumps are caused either by controllable, uncontrollable external events or by its continuous evolution. The continuous evolution and these jumps in control loops are the origins from the most stringent real-time demands. With the necessity to launch more satellites, Brazilian National Institute for Space Research (INPE) has been carrying out research on modeling and verifying hybrid systems, of which its main focus is to obtain a better balance between dependability, schedule, and cost. We are attempting to use Object Management Group (OMG) specifications to model discrete events. We are focusing mainly in Modelica (with some degree of Scicoslab) to model continuous dynamics. Another concern addressed by this, INPE, research is to be independent from commercial tools, establishing itself on open source software. This paper presents an approach to implement Model-Driven Architecture in hybrid systems based on vendor neutral specifications. It shows how the models are defined, traced and used, as well as a set of tools for this. SysML (Systems Modeling Language), and MARTE (Modeling and Analysis of Real-Time Embedded Systems) allowed us to define a Computation Independent Model focused mainly on high-level structure and behavior (state oriented). At the end, a case study is presented (inverted pendulum). From this case study, we have concluded that the proposed approach can complement uncovered topics in current research applied to hybrid systems development and maintenance.