Runway: Manufacturing AI System with Real-Time PLC Data

An anomaly detection model was created to enhance the quality and productivity of glass and ceramic processes used in displays. However, the absence of an AI operating system has led to increased development and maintenance costs due to the need for continuous one-off developments. Specifically, the company faces limitations in advancing AI model performance by not utilizing various features and real-time high-resolution data from PLCs at the production site.

Legacy systems on the manufacturing floor, including Historian, were connected to the edge, enabling AI model inference on on-premises servers to ensure data security and system connectivity. Specialized AI models were developed, and the retraining-redeployment process was automated, leveraging Rockwell's domain expertise in manufacturing and automation.

Maintenance costs were reduced by approximately 20%, and downtime due to equipment anomalies was minimized by up to 8% through a proactive, data-driven approach. Real-time AI-based inferences were delivered to PLC interfaces on the manufacturing floor, driving product quality improvement activities based on AI predictions.