A Semi-Automatic Approach for Asset Administration Shell Creation from Heterogeneous Data

Abstract

Heterogeneous data sources and inconsistent interfaces between different engineering disciplines pose a tremendous challenge in creating a virtual representation of a real-world product or plant. To enable the interoperability between all partners in the value chain, the Plattform I4.0, consisting of German academia and industry, proposed the Asset Administration Shell (AAS), a standardized concept of digitally representing of the asset. Unfortunately, the creation of a product or process AAS implies huge manual effort in collecting and structuring information pieces from diverse engineering and operation data by hand. This work proposes a semi-automatic approach to extract engineering information from documents of the data types PDF, STP, XML, XML/AML, URDF and a method to execute AAS creation and merging. We analyze the information structure in engineering files and map it to the meta-model of the AAS. We propose a Python implementation that semi-automatically generates AASs from engineering documents and merges them into an integrated AAS representing the asset. By using the demonstrator xPPU and two industrial use cases of gear assembly and robot control, we discuss the possibilities and limitations of our proposed approach.