The Research Of Model-Driven Modeling Specification For Cyber-Physical Systems

Cyber-Physical Systems(CPSs) are intelligent systems which integrate com-puting, communication and control processes. In CPSs. computation and physical processes effect each other to realize remote, reliable, real-time and collaborative interaction. CPSs are applied in many aspects of our daily life, especially in safety-critical systems, such as traffic control systems, automobiles, aerospace systems and so on. Thus, how to combine the discrete systems and continuous systems becomes a great challenge for CPSs.With the development of communication technology, computer technology and automatic control technology, communication based train control(CBTC) has be-come the newest railway signal system. CBTC is a safety-critical system which mainly contains ground equipment and vehicle equipment. CBTC is a typical cyber-physical system as its safe operation needs the collaboration of all sub-modules. Train Position Determination(TPD) system is a sub-system of CBTC which which requires superior security. TPD is also a hybrid system whose entities such as ve-locity, acceleration and travel distance evolve continuously during the process of positioning. The safe operation of train are closely related to the train’s speed, ac-celeration and position. Therefore, modeling the TPD system from the perspective of hybrid dynamics is very necessary and meaningful.At present, the specification in Model Driven Development(MDD)can only sup-port the modeling of discrete dynamics instead of continuous dynamics. The UML Profile for Modeling and Analysis of Real-Time and Embedded Systems(MARTE) add the capabilities to UML for MDD of RTES. To a certain degree, MARTE can be adopted in the modeling of CPS systems. MARTE provides a detailed time model and can support the unified modeling of discrete/continuous clocks and log-ical/chronometric clocks. Moreover, MARTE incorporates the Clock Constraint Specification Language(CCSL) for assisted modeling. However, the semi-formal property of MARTE makes it lack the detailed definition of formal syntax and se-mantics. In addition, the detailed description continuous behaviors of CPS has been not discussed. Thus, MARTE shoule be extended for the modeling of CPS.For this purpose, by analyzing the hybrid characteristics of CPSs, we extend MARTE with hybrid automata and hybrid transition system and propose a new modeling specification which is specific to CPS systems, that is HybridMARTE. The works of this paper are listed as below:1. We extend two meta-elements in MARTE:hybrid data types and expressions. In order to describe the continuous dynamics of variables, we extend data types to discrete variables, continuous variables, clock variables and event variables. Accordingly, there are four kinds of expressions in HybridMARTE:Boolean expression, Differential expression, clock constraints expression and action ex-pression.2. We provide a new unified clock model to support the modeling of both contin-uous clock and discrete clock.3. Based on hybrid transition system, a formal syntax and semantics of Hybrid-MARTE Class Diagram(HCD), HybridMARTE Sequence Digram(HSD) and HybridMARTE Statechart(HSC)are given.4. Based on the tool tMDA(Trustable Model Driven Architecture), we develop a new tool named Hybrid tMDA(Hybrid Trustable Model Driven Architecture)to support the modeling, simulation and verification of HybridMARTE models.In the end of this paper, we model the TPD with Hybrid MARTE. By analyzing the simulation and verification result, we demonstrate the specific advantage of Hvbrid MARTE in the domain of CPSs.