Knowledge Inference Over Web 3.0 for Intelligent Fault Diagnosis in Industrial Internet of Things

Collaboration through knowledge sharing is critical for the success of intelligent fault diagnosis in a complex Industrial Internet of Things (IIoT) system that comprises various interconnected subsystems. However, since the subsystems of an IIoT system may be owned and operated by different stakeholders, sharing fault diagnosis knowledge while preserving data security and privacy is challenging. While decentralized data exchange has been proposed for cyber-physical systems and digital twins based on the Web 3.0 paradigm, decentralized knowledge sharing in knowledge-based intelligent fault diagnosis is less investigated. To address this research gap, we propose a Web 3.0 application for collaborative knowledge-based intelligent fault diagnosis using blockchain-empowered decentralized knowledge inference (BDKI). Our proposed mechanism enables workers to self-evaluate their ability to contribute to the knowledge inference with their local knowledge graphs. The knowledge-sharing requestor can then choose a worker with the best evaluation result and initiate collaborative training. To demonstrate the efficiency and effectiveness of BDKI, we evaluate it using well-known datasets. Results show that BDKI delivers a favorable inference model with higher overall accuracy and less training effort compared to inference models trained using conventional knowledge inference with random training sequences.