Correct and Cost-efficient Development of Control Systems Based on Model Integration

The increasing demands on functionality and quality of the software in control systems are in contradiction with the increasingly shorter deadlines for its development and insufficient budget. The majority of software errors occurs in the early phases of its development, but is detected in later phases of testing. Later identified errors lead to extremely high costs for their removal. To overcome these drawbacks, new approaches for development of real-time software are necessary. The main objective of this paper is to present a framework for model-driven development of control systems based on UML and reinforced with additional opportunities for an early analysis and verification of the models. For this purpose an approach for integrating UML state machine and timed automata models has been developed and used. The integration is achieved through model transformation, based on meta-models and the transformation language ATL, plug-in in Eclipse platform. The proposed approach is illustrated with a simple example for discrete-event control of air compressor system. Finally, some conclusions are made.