Deep Learning‐Assisted Fingerprint‐Inspired Flexible Pressure Sensor for Tension Monitoring in Carbon Fiber Production

Abstract Carbon fiber has become a key emerging material in fields such as aerospace, wind power generation, and new energy vehicles. However, the current mass production of carbon fiber is limited by the challenges in controlling tension stability across wide‐width carbon fiber tow arrays, which restricts the performance and quality stability of the final product. Inspired by ancient textile workers who used their fingers to feel the tension of fiber bundles, a flexible pressure sensor is fabricated via laser etching to mimic the structure of fingerprints for monitoring the tension of wide‐width carbon fiber bundle arrays. This sensor demonstrates high sensitivity (18.08 kPa −1 ) and a broad pressure range of 320 kPa, with a maximum measurable pressure of 550 kPa. By installing the sensor array on the surface of a tension roller, multi‐fiber tension detection is achieved with a sensitivity of 5.55 N −1 . Furthermore, an end‐to‐end tension anomaly classification convolutional neural network is developed and achieved a high classification accuracy of 96%. This sensor offers a practical solution for real‐time, intelligent tension monitoring in carbon fiber production, which helps address key technical challenges in large‐scale, high‐consistency manufacturing and provides a new application for advancing sensing technology in flexible electronics and smart manufacturing.