A Digital Twin and Knowledge Graph Fusion Framework for Industrial Intelligence in the Context of Industry 5.0 and Industrial <scp>IoT</scp>

ABSTRACT Digital Twins (DT) map physical states and predict operations, while Knowledge Graphs (KG) provide structured semantic representation and logical inference. Their separation, however, creates data silos, weak interpretability, and absent closed‐loop control. This paper proposes Quadruple‐Layer (Q‐Layer), a four‐layer DT–KG fusion architecture that integrates entity mapping, semantic association, real‐time state synchronization, and cognitive decision‐making to establish a closed perception–semantics–inference–decision loop. Four reusable fusion patterns are derived to address varying industrial demands. Experiments on a wind turbine testbed validate the framework by integrating multi‐source sensor data into semantic networks and driving closed‐loop diagnosis via a customized multi‐scale convolutional neural network with attention (MSCNN_Attention). In the context of Industry 5.0 and Industrial IoT, Q‐Layer delivers semantic interoperability and human‐centric cognitive control in high‐stakes domains such as wind power predictive maintenance, although its performance still depends on the completeness of the pre‐built domain ontology.