基于YOLOv8的轴承故障图像识别方法  

Bearing Fault Image Recognition Method Based on YOLOv8

作　　者：安广析 姜利坤 AN Guangxi;JIANG Likun(Shandong Shanshui Heavy Industry Co.,Ltd.,Jinan 250306,China;Advanced Materials Institute,Qilu University of Technology(Shandong Academy of Sciences),Jinan 250013,China)

机构地区：[1]山东山水重工有限公司,山东济南250306 [2]齐鲁工业大学(山东省科学院),新材料研究所,山东济南250013

出　　处：《微型电脑应用》2024年第11期217-221,共5页Microcomputer Applications

摘　　要：由于长时间运行和外部环境等因素的影响,轴承往往容易发生各种故障,如磨损、裂纹和松动等。传统的轴承故障检测方法往往依赖于人工目测或简单的图像处理技术,存在识别速度慢、准确率低等问题,无法满足工业生产中对实时性和精确性的要求。深度学习技术在目标检测技术领域取得了巨大的进展,基于YOLOv8模型并引入全注意力机制(GAM),提出一种新的轴承故障图像识别方法。在西储大学轴承数据集上进行损失函数的消融实验,并与YOLOv8、SSD模型和Faster-RCNN模型进行比较实验,验证了所使用YOLOv8的高效检测性能。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 processing 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 technology.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.

关 键 词：轴承 故障识别 YOLOv8 全注意力机制 一阶段检测 

分 类 号：TM303[电气工程—电机]