Research on Online Detection Method for Lateral Deviation of Steel Strip in Steel Production Line Based on Machine Vision

To improve production efficiency and reduce production costs, machine learning methods for target recognition of online slabs have become a solution. This article mainly studies how to use machine learning technology for target recognition of online slabs and correcting the recognition results. To obtain the position information of the slab in the real world, this paper studied the slab coordinate measurement algorithm based on the camera imaging principle. By studying object detection algorithms based on deep learning, a machine learning detection model was established by preprocessing image data using Gaussian filtering, feature extraction of key points, and model training. Using the hot rolling production line of a steel manufacturing facility as an experimental platform, a slab deviation recognition system was developed for the hot rolling heating furnace. The hardware system integrated industrial cameras and computer servers, while Python programming software was utilized for system development. The system has a recognition running time of less than 0.5 seconds, an offset recognition accuracy of 95 %, good precision, meets industrial production needs, reduces labor intensity, and improves production efficiency.