Knowledge-graph-enhanced disturbance control in manufacturing systems: a state-of-the-art review

In the context of Industry 5.0, intensified competition and dynamic uncertainty have made efficient control essential for ensuring quality, quantity, and timely delivery in discrete manufacturing systems. However, production processes are vulnerable to unpredictable disturbances with complex, interrelated root causes. This paper reviews the use of knowledge graphs in disturbance control within manufacturing systems to support knowledge sharing and reuse. It first classifies manufacturing disturbances based on factory physics, production bottlenecks, and production factors, considering precedence, impact, and root causes. Next, it examines the integration of knowledge graphs with manufacturing systems, including application scenarios, available datasets, modeling approaches, and Smart Question & Answer systems. The current progress in knowledge-graph-driven disturbance control is then explored, and a disturbance control model based on Physics-Informed Neural Networks (PINNs) is proposed. Finally, the paper summarizes existing research and outlines future directions.