Semantic Runtime Interface Description Based on Engineering Knowledge Petr Nov´ ak *,** Martin Melik-Merkumians *** Michael Steinegger *,*** Thomas Moser * Radek ˇ Sindel´ aˇ r * Alois Zoitl *,*** * Christian Doppler Laboratory for Software Engineering Integration for Flexible Automation Systems, Vienna University of Technology, AT-1040 Vienna, Austria; email: ({novak,moser,sindelar}@ifs.tuwien.ac.at, {melik-merkumians,steinegger,zoitl}@acin.tuwien.ac.at ). ** Department of Cybernetics, Faculty of Electrical Engineering, Czech Technical University in Prague, 121 35 Prague, Czech Republic. *** Automation and Control Institute, Vienna University of Technology, AT-1040 Vienna, Austria. Abstract: Design and integration of industrial automation systems require cooperation of different engineering disciplines and various engineering tools during both design and run time. The scope of this paper is to bridge the gap between design-time description and runtime integration in a semi-automated way. Furthermore, we address the problem of integrating legacy systems with no access to original engineering data. It is assumed that an interface between a real industrial system and software automation tools is OPC Unified Architecture (UA) and we propose to derive the plant knowledge from the OPC UA tag list. Since it is not possible to solve this task in general, we require the tag list to adopt the naming convention of IEC 81346, which is widespread in large-scale systems. This tag list is parsed and the engineering knowledge is derived and stored in the Engineering Knowledge Base. Consequently, the plant knowledge is mapped to knowledge related to software automation tools, such as simulators. The proposed methodology is evaluated on a real-life example dealing with a laboratory tank model. Keywords: Automation, Integration, Knowledge-based systems, Data acquisition, OPC UA, IEC 81346, Computer Simulation, Software tools 1. INTRODUCTION Complex automation systems are designed and maintained by engineers originating from diverse engineering disci- plines. Each engineer uses a large variety of engineering tools providing many functionalities in specific domains. For example, electrical engineers apply tools to design and analyze electrical circuits, control engineers utilize dynamic simulations or optimizers, and process engineers use technology plans such as piping and instrumentation diagrams (P&ID) or tools to manage OPC tags. Although many of these tools are very powerful, their integration still remains a challenging problem, especially if various engineering domains have to be bridged. This paper is focused on integration of legacy automation systems and tools, i.e., the systems or tools implemented formerly. We assume that these systems work well and the industrial customer is satisfied with them. Any change of these systems would cause high development time and costs, therefore, this paper deals with deriving knowledge about such systems and integrating them as-they-are. Whereas design and integration methodologies starting from scratch are based on knowledge gathered at the early automation project stage; in the case of legacy systems the methodology is strongly different. First, the knowledge about a real plant and automation tools must be derived from feasible sources or tools. Consequently, the knowledge must be stored in an appropriate knowledge base. Finally, the mapping between concepts and interfaces from different domains representing the same entities is created. This paper proposes to derive such a mapping semi-automatically, but due to security reasons, the results have to be confirmed by integration engineers. Since the communication in industrial automation practice is expected to be mostly based on OPC UA, the paper proposes to derive plant knowledge semi-automatically out of OPC UA tag lists being in compliance with IEC 81346, ref. IEC 81346. Adopting the naming convention defined by this standard is the main assumption of the proposed methodology. As the standard is widely used in power plants and in many other large-scale industrial cases, it does not pose a significant limitation. The IEC 81346 tag list is not only a textual list, since each tag name has a well- defined structure, representing a semantic content. The tag list is processed to get an initial portion of knowledge required for integration. In order to support efficient and semi-automated algorithms, we propose to use the so- called Engineering Knowledge Base (EKB), which was presented and evaluated by Moser and Biffl (2010). Proceedings of the 14th IFAC Symposium on Information Control Problems in Manufacturing Bucharest, Romania, May 23-25, 2012 978-3-902661-98-2/12/$20.00 © 2012 IFAC 805 10.3182/20120523-3-RO-2023.00176