基于深度学习的滚动轴承剩余使用寿命预测  

Prediction of remaining useful life of rolling bearing based on deep learning

作　　者：万晓凡 封士瑞 张营[1] WAN Xiaofan;FENG Shirui;ZHANG Ying(College of Automotive and Transportation Engineering,Nanjing Forestry University,Nanjing 210037,China)

机构地区：[1]南京林业大学汽车与交通工程学院,南京210037

出　　处：《智能计算机与应用》2024年第9期125-130,共6页Intelligent Computer and Applications

摘　　要：滚动轴承是机械传动设备的“关节点”,对其进行剩余使用寿命预测对机械设备安全生产和维护有重要意义,本文提出一种基于粒子群优化算法配合深度学习的滚动轴承剩余使用寿命预测模型。首先,对滚动轴承振动信号进行时域、频域和时频域特征提取,利用单调性和鲁棒性筛选出能够反映轴承退化过程的敏感特征;其次,基于高斯混合模型提取健康因子,解决单一特征指标无法有效反映退化趋势的问题;最后,将粒子群优化后的网络结构参数输入模型中进行轴承的剩余使用寿命预测,通过两组数据集的预测结果比较发现,粒子群优化后双向长短时记忆神经网络模型预测精度比双向长短时记忆神经网络的模型提高约10.6%和24.7%。Rolling bearings are the"joint points"of mechanical transmission equipment,and the prediction of their remaining service life is of great significance to the safe production and maintenance of mechanical equipment.In this paper,a prediction model for the remaining service life of rolling bearings based on deep learning is proposed.Firstly,the time-domain,frequencydomain and time-frequency domain features of the rolling bearing vibration signal were extracted,and the sensitive features that could reflect the bearing degradation process were screened out by using monotonicity and robustness.Secondly,the health factors were extracted based on the Gaussian mixture model to solve the problem that a single characteristic index could not effectively reflect the degradation trend.Finally,the network structure parameters of the particle swarm optimization are input into the model to predict the remaining service life of the bearing,and the prediction accuracy of the two-way long short-term memory neural network model after particle swarm optimization is about 10.6%and 24.7%higher than that of the two-way long short-term memory neural network model.

关 键 词：滚动轴承 剩余使用寿命 深度学习 

分 类 号：TB389.1[一般工业技术—材料科学与工程]