基于时频融合深度网络的矿用钻机轴承故障诊断
Bearing Fault Diagnosis of Mining Drilling Rig with Time-frequency-fused Deep Network
邹筱瑜 1孙国庆 2王忠宾 1潘杰 3刘新华 1李鑫1
作者信息
- 1. 中国矿业大学机电工程学院,徐州,221116;智能采矿装备技术全国重点实验室,徐州,221116
- 2. 中国矿业大学机电工程学院,徐州,221116
- 3. 中国矿业大学信息与控制工程学院,徐州,221116
- 折叠
摘要
针对矿用钻机低速重载工作特性导致其轴承故障特征微弱、噪声大的问题,考虑单一模态下故障诊断的局限性,提出了一种基于时频融合深度网络的矿用钻机轴承故障诊断方法,对时域和时频域两种模态特征进行联合提取与分析.所设计的诊断网络在不同模块中区分性地嵌入不同注意力机制,实现多维度关键故障特征提取.最后通过钻机实验台数据集和凯斯西储大学轴承数据集进行验证.结果表明:所提方法能自动提取丰富的钻机轴承故障特征,比仅在时域或时频域分析具有更高的准确率和抗噪能力.
Abstract
To solve the problems of weak and noisy bearing fault features caused by the low-speed and heavy-load operating characteristics of mining drilling rigs,a fault diagnosis method was proposed for mining rig bearings,named time-frequency-fused deep network.It considered the limitations of fault diagnosis with single modality,and then jointly characterizes two modal features of the time do-main and time-frequency domain.The designed diagnostic network differentially embeded specific at-tention mechanism in different modules to extract multi-dimensional key fault features.Finally,the proposed method was validated on the experimental equipment and the Case Western Reserve Univer-sity bearing dataset.The results show that the proposed method may automatically extract sufficient fault features combining two domains.It has higher accuracy and noise immunity than those with a single domain.
关键词
矿用钻机轴承/故障诊断/时频融合/注意力机制/空洞卷积Key words
bearings of mining drilling rig/fault diagnosis/time-frequency fusion/attention mechanism/dilated convolution引用本文复制引用
基金项目
国家自然科学基金(62273349)
国家自然科学基金(62176258)
国家重点研发计划(2020YFB1314200)
中央高校基本科研业务费(2021YCPY0111)
江苏省高校优势学科建设工程(PAPD)()
出版年
2024
国家科技期刊平台
NSTL
万方数据