摘要
针对刀具剩余寿命预测问题,提出了一种将一维卷积自编码器(One-dimensional convolutional auto encoder,1DCAE)和残差双向门控循环单元(Residual bidirectional gated recurrent unit,RBGRU)相结合的预测方法。通过1DCAE连续卷积池化和反卷积上采样方法获取工况信号的深层特征,并将其与分段后的原始信号融合后作为刀具剩余寿命的表征;同时结合残差网络的思想对双向门控循环单元(Bidirectional gated recurrent unit,BiGRU)的结构进行改进以增强对时序特征的捕获能力。实验结果表明,该方法比其他算法具有更高的预测精度。
Focusing on the remaining useful life prediction of cutting tool,a method combining one-dimensional convolutional auto encoder(1DCAE)with residual bidirectional gated recurrent unit(RBGRU)is proposed.The underlying features of the working condition data are obtained by using 1DCAE continuous convolution pooling and deconvolution upsampling methods,and fused with the segmented original signal.And the fused data is used to characterize the remaining life of tool.Meanwhile,the structure of the bidirectional gated recurrent unit(BiGRU)is improved by combining the idea of residual network to enhance the capability of capturing the timing features.The experimental results show that the prediction accuracy of the present method is better than that via the other algorithms.
作者
周学良
潘晓明
吴瑶
ZHOU Xueliang;PAN Xiaoming;WU Yao(School of Mechanical Engineering,Hubei University of Automotive Technology,Shiyan 442002,Hubei,China)
出处
《机械科学与技术》
北大核心
2025年第5期806-813,共8页
Mechanical Science and Technology for Aerospace Engineering
基金
湖北省高等学校优秀中青年科技创新团队计划(T2020018)
第64批中国博士后科学基金项目(2018M6409120)。
关键词
刀具
剩余寿命预测
卷积自编码器
残差门控循环单元
特征融合
cutting tool
remaining useful life prediction
convolutional auto encoder
residual gated recurrent unit
feature fusion
