摘要
针对在复杂运行环境下,轴承故障振动信号受强噪声干扰而难以进行特征提取的问题,以及高维特征虽然包含更多运行状态信息但是存在"维数灾难"的问题,提出了基于ITD与MLE-LLTSA的故障诊断方法。首先利用ITD算法对原始故障信号进行分解得到一系列的固有旋转分量(PRC),根据峭度准则与相关系数,筛选出含故障信息丰富的PRC进行相加重构得到新的数据集;提取重构后数据集的时域、频域等多个特征构建多域特征向量集;利用MLE与LLTSA相结合得到高维多域特征集的低维本征流形,最后训练得到基于极限学习机(ELM)轴承故障诊断模型。采用公开轴承数据集来进行测试实验,该模型的故障识别率达到了99.20%,具有较高的分类精度。实验结果表明,所提方法能够有效提取出故障振动信号并减少高维特征所产生的冗余问题,实现对轴承运行状态的有效识别。
In order to solve the problem that the vibration signal of bearing fault is disturbed by strong noise and difficult to extract the feature,and the problem that the high-dimensional feature contains more running state information but could cause"dimension disaster",a fault diagnosis method based on ITD and mle-lltsa is proposed.In this method,the original fault signal is decomposed by ITD algorithm to obtain a series of PRC.According to kurtosis criterion and correlation coefficient,the PRC with rich fault information is selected to add and reconstruct to get a new data set.After the reconstruction,multiple features such as time domain and frequency domain are extracted to build a multi-domain feature set.By combining MLE with LLTSA,the intrinsic manifold of high dimensional and multi-domain feature set is obtained.Finally,the bearing fault diagnosis model based on the ELM is trained.The public bearing data set is used to test the model.The fault recognition rate of the model reaches 99.20%,which has a high classification accuracy.The experimental results show that the method can effectively extract the fault vibration signal and reduce the redundancy caused by high-dimensional features,and realize the effective recognition of the bearing running state.
作者
肖洁
黎敬涛
邓超
罗志刚
林华峰
Xiao Jie;Li Jingtao;Deng Chao;Luo Zhigang;Lin Huafeng(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,China;Electrical Engineering Co.,Ltd.of China Railway 12th Bureau Group,Tianjin 300000,China)
出处
《电子测量技术》
2020年第8期183-188,共6页
Electronic Measurement Technology
作者简介
肖洁,硕士研究生,主要研究方向为机械设备故障诊断。E-mail:785228291@qq.com;通信作者:黎敬涛,副教授,硕士生导师。E-mail:179015023@qq.com
