Data-driven ontology generation and evolution towards intelligent service in manufacturing systems

Abstract

Abstract To support intelligent manufacturing, providing a unified production data view by integrating distributed data collected by different enterprise information systems is critical. Because various information systems are often heterogeneous, ontology is widely adopted to present a global reference view for data integration. However, construction and maintenance of these ontologies is difficult because of the heterogeneity and dynamism of these large-scale data. In this paper, with the objective of intelligent manufacturing application implementation, we propose a comprehensive ontology generation and evolution method that automatically abstracts ontology from raw production data and dynamically adjusts the ontology in accordance with changes in the manufacturing data environment. The proposed method comprises four phases: data extraction, ontology construction, ontology connection, and ontology evolution. In the first phase, data from different sources are mapped to data entities to form a unified data structure. In the second phase, an initial ontology is generated via instance-driven ontology construction. In the third phase, to support intelligent manufacturing, the initial ontologies are organised in terms of the dimensions of the various business elements, such as stuff, machine, product, process, and scenarios. In the fourth phase, rules regarding ontology restrictions are formulated to realise ontology evolution that respond to changes in the manufacturing environment. To verify the efficacy of the proposed method, a prototype was implemented with real data from a manufacturing factory, in which the constructed ontology was used as the metadata of product data in intelligent manufacturing.