Validation of AutomationML Models

Abstract

The planning, implementation and commissioning of industrial production plants is a complex endeavour which requires the collaboration of a variety of engineering disciplines (e.g. electrical engineering, mechanical engineering, etc.). Due to industry specific software requirements, highly specialized software is used in the involved branches. Thisresults in proprietary data formats without possibility of data exchange between different engineering disciplines. As this is a time consuming and error prone task with the risk of misinterpretation or lost data this issue was identified as reason for delayed projects.Therefore leading companies in the field of automation triggered the creation of AutomationML as a solution to the problem. It uses the hierarchical, neutral, XML based data format Computer Aided Engineering eXchange (CAEX) as its top-level format and adds additional restrictions and extensions.In order to ensure efficient and error-free data exchange, it must be ensured that the data conforms to the underlying data format. So far there has been no way for AutomationML models to validate their conformity with respect to restrictions and extensions invented by the AML specification.This thesis implements a validation framework for AutomationML models using tools from the field of model-based software development. The already existing validation regarding conformity to the CAEX data format is used as basis. In addition, the extensions and restrictions defined by the AML specification are formalized as part of theimplementation.The implemented validation framework examines AutomationML models in a structured process for conformity with regard to the AML specification and reports deviations. For evaluation purposes test models which deliberately deviate from the AutomationML specification were created. Based on these faulty models the correct detection of deviations is checked. In addition AML models conforming to the AutomationML were createdto ensure that correct models are recognized as such.