摘要
针对传统滚动轴承故障诊断方法准确率偏低和故障特征难以提取的问题,提出了一种基于SDP图像和MobilenetV2的滚动轴承故障诊断方法。将经去噪处理后的滚动轴承振动信号转化为SDP图像,并输入到MobilenetV2网络中自适应地提取故障特征和分类,诊断出具体的故障类型。试验表明,在适量的训练样本下,所提方法的诊断准确率能达到98.2%。相比于其它传统深度学习方法,所提方法在诊断正确率和稳定性方面具有一定优势。在原始轴承振动信号中加入5 dB的高斯白噪声后,故障识别准确率仍能达到94.4%,证明了所提出诊断方法具有一定的抗噪性能和泛化能力。
Aiming at the low accuracy of traditional rolling bearing fault diagnosis methods and the difficulty in extracting fault features, a rolling bearing fault diagnosis method based on SDP image and MobilenetV2 was proposed.The denoised vibration signal of rolling bearing is converted into SDP image, and input into MobilenetV2 network to extract fault features and classification adaptively, and diagnose specific fault types.Experiments show that the diagnostic accuracy of the proposed method can reach 98.2% with a suitable amount of training samples.Compared with other traditional deep learning methods, the proposed method has certain advantages in diagnostic accuracy and stability.After adding 5 dB white Gaussian noise to the original bearing vibration signal, the fault identification accuracy can still reach 94.4%,which proves that the proposed diagnosis method has certain anti-noise performance and generalization ability.
作者
刘昕宇
姜长泓
王其铭
张同晖
LIU Xin-yu;JIANG Chang-hong;WANG Qi-ming;ZHANG Tong-hui(School of Electrical and Electronic Engineering,Changchun University of Technology,Changchun 130012,China)
出处
《组合机床与自动化加工技术》
北大核心
2023年第2期178-182,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
吉林省科技发展计划项目(20220203041SF)。