DF-MKG-DTS: An integrated knowledge graph-based data representation method for digital twin shop-floors

Digital Twin Shop-floor (DTS) systems require effective representation and integration of heterogeneous manufacturing data to support real-time decision-making. Existing approaches often suffer from limited semantic connectivity and insufficient support for dynamic data updates, restricting their applicability in complex manufacturing environments. To address these challenges, this paper proposes a DF-MKG-DTS framework that integrates a Manufacturing Knowledge Graph (MKG) into a Data Fabric (DF) architecture. The framework enables unified representation, semantic association, and dynamic updating of multi-source shop-floor data by modeling manufacturing entities, relationships, and temporal attributes as a structured Knowledge Graph, thereby improving fault localization accuracy. Experimental results demonstrate that the data association rate reaches 99.5% before knowledge fusion and achieves full association after fusion, validating the effectiveness of the proposed approach. For small-scale temperature faults, the proposed framework achieves 92.6% accuracy, an improvement of 10.2% over manual inspection, and maintained 89.8% accuracy for large-scale, complex process combination faults while reducing average processing time by up to 40%. These results indicate that DF-MKG-DTS not only alleviates data isolation and enhances data utilization but also provides robust, scalable, and efficient fault diagnosis support, highlighting its practical value for Digital Twin Shop-floor applications.