欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
为提取高速列车转向架关键部件振动信号的特征,提出基于深度学习（Deep Learning）的高速列车转向架故障识别新方法。以转向架关键部件非全拆单工况故障信号为研究对象,对故障信号进行离散傅里叶变换,然后依据深度学习的降噪自动编解码过程对故障的频域信号进行特征学习,并以此特征作为BP神经网络的输入实现转向架故障信号的识别。实验结果表明：在不同速度下,所提方法对转向架关键部件非全拆故障识别正确率能达到100%,表明了该方法的有效性。
英文摘要：
To extract the features of vibration signals of high- speed train bogie,a novel approach,which is based on Deep Network,was proposed to recognize the faults of high- speed train bogie. Fault signals were obtained from non- whole single fault of key components of bogie. The discrete fourier trandform was conducted to the fault signals and then a denoising autoencoders was presented to abstract features of vibration signals. Based on the deep learning concept,a deep network was build to recognize the faults of high- speed train bogie. The experimental results show that the recognition rate is 100% for non- whole fault of key components of bogie at the different speed,which verified the effectiveness of the proposed method.
同期刊论文项目
同项目期刊论文
期刊信息
位置：