Cap ply strips, the nylon cord reinforcement layers over steel belts, are essential for tire durability and belt edge stability. Pirelli presents an in-line cord-break detection system for cap ply strip production lines, using a multithreaded deep-learning vision pipeline with six stations and cameras. Sequential frames provide pattern features, inter-frame differences and optical-flow vectors. The model outputs motion/stop probabilities, and a process-specific decision rule with experimental thresholds triggers alarms. Trained on ~80,000 images and running on an edge AI computer at line speed, the patented system is robust to vibration, illumination changes and different cord types.

• Designing a multithreaded, edge-deployed vision stack (producer(camera)/AI inference/GUI threads, decision-time optimization)
• Extracting pattern features, inter-frame differences and optical-flow vectors; training a custom DL model (~80k images) to generalize across cord types
• From model to decision: crafting a process-specific rule using probability scores and experimental thresholds to deliver reliable alarms/actions
• Hardware selection criteria: camera, lens and industrial edge-AI computer
• Programming language choices by component