摘要
为了进一步提高对机床不同故障的分类准确率,设计了一种深度残差网络。通过对机床振动试验台信号预处理优化网络结果,并进行故障诊断对比分析。研究结果表明:当行数比列数更少时,随着两者差异增加,模型分类准确率显著降低;当行数超过列数后,模型达到了更高的分类准确率并保持相对稳定的状态。CNN网络比浅层模型LeNet表现出了更强识别性能。ShortCut结构具备明显优越性,有助于网络具备更强识别能力。该研究有助于提高减速器的运行效率,可将其拓宽到其他机械传动领域,具有很好的应用价值。
In order to improve the classification accuracy of different machine tool faults,a deep residual network is designed.The network results of machine tool vibration test bench signal preprocessing were optimized and the fault diagnosis was analyzed.The results show that when the number of rows is less than the number of columns,the classification accuracy of the model decreases significantly with the increase of the diference between the two.When the number of rows exceeds the number of columns,the model achieves a higher classification accuracy and maintains a relatively stable state.The CNN network shows stronger recognition performance than the shallow model LeNet.The ShortCut structure has an obvious advantage,which helps the network to have stronger identification capability.The research is helpful to improve the running efficiency of the reducer and broaden it to other fields of mechanical transmission,which has good application value.
作者
李彩霞
Li Caixia(College of Information Engineering,Henan Vocational College of Industry and Trade,Zhengzhou Henan 451191,China)
出处
《机械管理开发》
2024年第1期1-2,6,共3页
Mechanical Management and Development
基金
河南省重点研发与推广专项项目(202102210177)。
关键词
机床
残差网络
故障诊断
振动信号
shield cutter head
performance evaluation
t-distributed random neighborhood embedding
signal dimension reduction
