Due to long-term operation and external environmental factors,bearings are prone to various faults such as wear,cracks and looseness.Traditional methods for bearing fault detection rely on manual visual inspection or simple image process-ing techniques,resulting in slow recognition speed and low accuracy,which can not meet the requirements for real-time and precision in industrial production.Deep learning technology makes significant advancements in the field of object detection tech-nology.This paper introduces a global attention mechanism(GAM)based on the YOLOv8 model and proposes a new method for recognizing bearing faults.Experiments are conducted on the bearing dataset of Western Reserve University,compared with YOLOv8,SSD model and Faster-RCNN model,the efficient detection performance of YOLOv8 is verified.
维普
万方数据