dc.description.abstract
Production systems are complex structures. The growing need for customized mass-goods and decreasing life cycles of consumer goods imposes the demand for flexible manufacturing systems and frequent changes to deployed plants. Different disciplines are involved in their creation. More and more functionalities are covered by automatic control software. Hence, its share in terms of costs and development time is ever increasing. However, in prevalent, serialized development workflows, automatic control development is one of the last disciplines involved. This imposes pressure regarding functionality, costs, and also commissioning time on automatic control engineers. Model-based engineering is a promising approach to improve this situation as it enables and fosters the collaboration of all involved disciplines. The aim of this thesis is to introduce a new multi-disciplinary approach for the specification, development, and validation of production systems. The core of the proposed approach is the modelling infrastructure, which is inspired by the Model Driven Architecture (MDA). A set of models, each for different engineering aspects, is linked to provide coherent, multi-disciplinary data that is relevant for the implementation of production systems. Functional modules, so called Automation Components, are introduced. These are the building blocks of automated production systems. Furthermore, they facilitate the specification and implementation process, as the re-use of Automation Components reduces the engineering effort. Furthermore, once specified and implemented components usually have higher quality. The proposed workflow targets the simultaneous system specification in all involved disciplines. Currently used engineering tools, such as CAD tools or PLC programming tools, shall be integrated and model-transformation facilities are used to extract engineering data. The availability of a comprehensive specification early in the engineering workflow allows the parallelized implementation in the involved disciplines (e.g. mechanical construction, programming of the automatic control system). Hence, the overall development and implementation time can be reduced. For further reconfiguration activities it is important to keep the specification of the automated system accurate and coherent with the actual implementation. The hierarchical aggregation of Automation Components, with clearly defined interfaces and encapsulate behaviour, support this task, as changes are locally confined. The direct inclusion of plant behaviour simulation in the specification and development workflow is an innovation. This allows also to keep the simulation model of the automated plant coherent with the actual (or planned) plant configuration. Implementation can already be started without the physical availability of the controlled plant. The simulation environment provides a virtual representation of the plant and allows the validation of automatic control applications (i.e. virtual commissioning). The simulation framework is based on the IEC 61499 compliant runtime environment FORTE. Its event based execution semantics is suitable for discrete event simulation. Also the modular aggregation of functionality-based on Function Blocks-facilitates the modeling of modular plant models (with the same structure as the Automation Components). The automatic control application and the plant simulation application are executed on the same automatic control system (one or multiple automatic control devices). First, coupling plant simulation and automatic control is possible without communication overhead. Second, the execution of the automatic control application is not restricted, as it is the same type of runtime environment they are deployed to later in the field. Different scenarios, such as full simulation, hybrid simulation, inclusion of external simulation tools, and finally operation and testing, support automatic control engineers during various phases of the automatic control development. Finally, test cases from three different domains are selected. Discrete manufacturing plants, robotics applications, and process technology are covered. Each domain has unique requirments that have to be fulfilled by the proposed engineering process. For the evaluation of the proposed approach all relevant elements of the workflow are implemented. The feasibility of an integrated engineering and simulation workflow is validated with the help of these three test cases. Different models and interfaces (between the disciplines) are maintained through the collaboration of experts from multiple disciplines. Automatic control engineers gain access to the well-suited validation tool of discrete event simulation with little additional specification effort by reusing specification data for the automatic generation of simulation models. This helps to reduce the unproductive, and thus costly, commissioning and rampup time. With the establishment of comprehensive libraries of Automation Components the engineering effort, across all involved disciplines, is significantly reduced. Hence, the engineering cycles for the creation of production systems or their adaptation to meet new requirements are shortened.
en