Knowledge graph based OPC UA information model automatic construction method for heterogeneous devices integration
Xiaobin Li,Shucheng Zhang,Pei Jiang,Mikun Deng,Xi Vincent Wang,Chao Yin
DOI: https://doi.org/10.1016/j.rcim.2024.102736
IF: 10.103
2024-08-01
Robotics and Computer-Integrated Manufacturing
Abstract:Emergent manufacturing paradigms, including ubiquitous manufacturing, social manufacturing, and agile manufacturing, facilitate the advancement of the manufacturing industry. However, these paradigms suffer from the difficulty of interoperability due to the heterogeneity of field devices and diverse communication protocols. The OLE for Process Control Unified Architecture (OPC UA) technique, which is featured with cross-platform, is considered as the pivotal technology for addressing this issue. However, there still exist a large number of heterogeneous devices which do not support OPC UA protocol. The communication protocols of these heterogeneous devices need to be converted into OPC UA based on information model, nevertheless, the manual construction of OPC UA information model is cumbersome and inefficient. In this paper, an integrated architecture based on reasoning over the OPC UA information model is proposed to realize rapid integration of heterogeneous devices to achieve interoperability. To realize the proposed integrated architecture, an automatic device information model construction method is developed simultaneously. The method first identifies the type of the newly accessed device based on the character-level TextCNN (CTCNN) model, which utilizes the word sequence extracted from the corresponding device data frame as input. Subsequently an open-world knowledge completion model is adopted to link the unseen entities in the device data frame to the information model knowledge graph(KG) to support the information model automatic construction for unknown device. For evaluation purpose, two industrial datasets are constructed and the results demonstrate the feasibility and effectiveness of the proposed method.
robotics,computer science, interdisciplinary applications,engineering, manufacturing