【摘要】 轴承作为机械系统的重要组成部分，其动态特性分析及缺陷检测研究具有重要意义。本文以滚动轴承为对象，进行动力学特性分析及缺陷检测方法研究。建立纯刚体滚动轴承模型，定量分析不同工况对轴承内部构件动力学特性的影响，进一步基于多项式函数对各工况下的最大应力和最大变形量进行拟合，构建不同工况下轴承最大应力与变形量的预测模型；针对检测环境中面临的其他金属零件或同类轴承零件的不利影响，采用深度学习技术提出基于YOLOv5的轴承缺陷检测网络模型。实验结果表明，网络模型对轴承的擦伤缺陷、凹槽缺陷和划痕缺陷检测的AP分别可以达到94.51%,92.56%和89.47%，总体检测精度mAP为92.18%。更多还原

【Abstract】 As an essential component of mechanical systems, the dynamic characteristics analysis and defect detection of bearings are of great significance. In this study, rolling bearings are taken as the research object for their dynamic characteristic analysis and defect detection method. A purely rigid rolling bearing model is established to quantitatively analyze the influence of different working conditions on the dynamics of internal components. Polynomial functions are then used to fit the maximum stress and maximum deformation under various conditions, constructing a predictive model for the maximum stress and deformation of bearings under different working conditions. In order to solve the adverse effects of other metal parts or similar bearing parts in the detection environment, a bearing defect detection network model based on YOLOv5 is proposed using deep learning technology. The experimental results show that the network model achieves AP values of 94. 51%, 92. 56%, and 89. 47% for scratches defects, groove defects, and scratch defects, respectively, with an overall detection accuracy of mAP 92. 18%.更多还原