Knowledge graph-enhanced fault diagnosis: a bibliometric review of AI applications in sensor management (1998–2024)

Abstract Background Sensor fault management is critical for intelligent systems in energy, healthcare, and smart cities, where faults pose economic and safety risks. Despite AI and IoT advances, challenges remain in fault correlation analysis, model interpretability, and interdisciplinary collaboration. This study uses Socio-Technical Systems (STS) theory and bibliometric methods to address these gaps. Methods A bibliometric analysis of 1,495 Web of Science articles (1998–2024) was conducted via CiteSpace and VOSviewer, mapping trends, collaboration networks, and keyword evolution. A PRISMA-guided systematic review synthesized insights within an STS framework, proposing a knowledge graph approach for fault management. Results Publications grew exponentially (13.58% annual, R 2 = 0.9797), led by China (35.34%) and the U.S. (18.27%). Key themes included fault diagnosis, wireless sensor networks, and predictive maintenance, with recent focus on lithium-ion batteries. A graph neural network-based knowledge graph (a structured representation of fault relationships using AI techniques) achieved 95% diagnostic accuracy. Conclusion This study highlights the Socio-technical development of sensor fault management and knowledge graphs’ potential to enhance reliability. Future research should focus on scalable algorithms and interdisciplinary collaboration for IoT intelligent maintenance.