摘要
对轴承振动信号的累计特征进行提取,在此基础上提出基于循环卷积神经网络(Recurrent Convolutional Neural Network,RCNN)的滚动轴承剩余使用寿命(Remaining Useful Life,RUL)预测模型。该预测模型通过构建循环卷积层增强神经网络对时间依赖性的学习,通过变分推理量化RUL预测中RCNN的不确定性。通过试验将基于RCNN的滚动轴承RUL预测模型与回归预测模型的预测结果相对比,验证该基于RCNN的预测模型的有效性。
The cumulative characteristics of bearing vibration signals are extracted,and a rolling bearing RUL prediction model based on the Recurrent Convolutional Neural Network(RCNN)is proposed on this basis.The prediction model enhances the learning of time dependence of neural networks by constructing cyclic convolutional layers and quantifies the uncertainty of RCNN in RUL prediction with variational reasoning.The effectiveness of the RCNN prediction model is verified when the prediction results of the RCNN based prediction model is compared with that of the regression prediction model in the experiments.
作者
潘冬伟
范志川
姬永波
项乔
PAN Dongwei;FAN Zhichuan;JI Yongbo;XIANG Qiao(IT Research Institute,Shanghai JiangnanChangxing Shipbuilding Co.,Ltd.,Shanghai 201913,China)
出处
《船舶与海洋工程》
2023年第5期78-85,共8页
Naval Architecture and Ocean Engineering
基金
工信部高技术船舶项目(CJ04N20)。
