Knowledge-Driven Framework for Online Monitoring and Fault Diagnosis in Manufacturing Lines

Manufacturing lines pose substantial challenges for operational safety, system reliability, and real-time monitoring. To address these challenges, this paper proposes a comprehensive framework for online monitoring and fault diagnosis in manufacturing lines, which aim to eliminate human presence from hazardous production environments through automation and intelligence. Drawing inspiration from mature online monitoring and fault diagnosis techniques in single manufacturing system, the proposed method accounts for the high-dimensional, multi-source, and heterogeneous nature of industrial data in complex manufacturing lines that contains a number of manufacturing systems. The framework incorporates causal discovery and complex network analysis for key parameter identification, and further integrates a knowledge graph to support semantic reasoning for fault diagnosis and emergency strategy recommendation. This approach enables accurate real-time status assessment, interpretable fault analysis, and adaptive decision-making in high-risk, highly automated production scenarios.