Application of KTA-KELM in Fault Diagnosis of Rolling Bearing

Zhuo Wang; Wenjun Zhao; Tao Ma; Zhijun Li; Bo Qin

doi:10.16578/j.issn.1004.2539.2019.06.030

您当前的位置：

首页 >

文章列表页 >

Application of KTA-KELM in Fault Diagnosis of Rolling Bearing

Development·Application | 更新时间：2022-10-01

- Application of KTA-KELM in Fault Diagnosis of Rolling Bearing
- Journal of Mechanical Transmission Vol. 43, Issue 6, Pages: 165-171(2019)
- 作者机构：
  
  1.内蒙古第一机械集团公司工艺研究所，内蒙古包头 014032
  2.特种车辆及其传动系统智能制造国家重点实验室，内蒙古包头 014032
  3.内蒙古科技大学机械工程学院，内蒙古包头 014010
- 作者简介：
- 基金信息：
- DOI：10.16578/j.issn.1004.2539.2019.06.030
  CLC：
扫描看全文
王卓,赵文军,马涛等.KTA-KELM在滚动轴承故障诊断中的应用[J].机械传动,2019,43(06):165-171.

Wang Zhuo,Zhao Wenjun,Ma Tao,et al.Application of KTA-KELM in Fault Diagnosis of Rolling Bearing[J].Journal of Mechanical Transmission,2019,43(06):165-171.
王卓,赵文军,马涛等.KTA-KELM在滚动轴承故障诊断中的应用[J].机械传动,2019,43(06):165-171. DOI： 10.16578/j.issn.1004.2539.2019.06.030.

Wang Zhuo,Zhao Wenjun,Ma Tao,et al.Application of KTA-KELM in Fault Diagnosis of Rolling Bearing[J].Journal of Mechanical Transmission,2019,43(06):165-171. DOI： 10.16578/j.issn.1004.2539.2019.06.030.

摘要

在数据驱动的滚动轴承状态辨识模型构建过程中，针对核极限学习机（Kernel Extreme Learning Machine，KELM）算法中高斯核函数的径向宽度参数,σ,选取不当极易造成模型分类精度差的问题，提出一种核排列优选核参数,σ,的滚动轴承状态辨识方法。首先，将测取滚动轴承振动信号经总体局部均值分解（Ensemble Local Mean Decomposition，ELMD）进行分解并计算其能量熵、排列熵来构建高维的特征向量集；然后，初始化核排列（Kernel Target Alignment，KTA）算法参数：最大核排列值,A,i,和核参数,σ,i,，通过判断核矩阵与理想目标矩阵间距离来调节不同的,A,i,和,σ,i,值，来获取两矩阵距离最短时所对应的,A,i,，此时核参数,σ,i,最优。最后，将上述滚动轴承的高维特征向量集作为输入通过KTA-KELM算法的学习，建立基于KTA-KELM的滚动轴承的状态辨识模型。仿真实验结果表明，与KELM、ELM算法相比，KTA-KELM算法将滚动轴承状态辨识的精度由92.5%和90%提高到98.75%，分别提高6.25%和8.75%。

Abstract

In the process of data-driven rolling bearing state identification model construction，the improper selection of the radial width parameter σ of the Gaussian kernel function in the Kernel Extreme Learning Machine（KELM）algorithm is very easy to cause poor classification accuracy. A method for identifying the state of rolling bearings based on the kernel arrangement preferred kernel parameter σ is proposed. Firstly，the rolling bearing vibration signal is decomposed by Ensemble Local Mean Decomposition（ELMD）and its energy entropy and permutation entropy are calculated to construct a high-dimensional eigenvector set. Then，the Kernel Target Alignment（KTA） parameters of maximum KTA value Ai and the kernel parameter σi are initialized, and the different kernel parameter values are adjusted by judging the distance between the kernel matrix and the ideal target matrix，so as to obtain the minimum corresponding maximum kernel arrangement value when the kernel matrix distance is obtained，and the kernel parameter at this time is optimal. Finally，the high-dimensional feature vector set of the above rolling bearing is used as input to learn the KTA-KELM algorithm, the state recognition model of rolling bearing is built based on KTA-KELM algorithm. The simulation results show that compared with KELM and ELM，the KTA-KELM algorithm improves the accuracy of rolling bearing state recognition from 92.5% and 90% to 98.75%， which is increase 6.25% and 8.75%.

关键词

滚动轴承核参数优化状态辨识分类精度

Keywords

Rolling bearingKernel parameter optimizationState identificationClassification precision

references

曹宏瑞，李亚敏，成玮，等. 局部损伤滚动轴承建模与转子系统振动仿真［J］. 振动、测试与诊断， 2014， 34（3）：549-552.

LIU J， SHAO Y， ZHU W D. A new model for the relationship between vibration characteristics caused by the time-varying contact stiffness of a deep groove ball bearing and defect sizes［J］. Journal of Tribology， 2015， 137（3）： 031101.

RUBINI R. Application of the envelope and wavelet transform analyses for the diagnosis of incipient faults in ball bearing［J］. Mechanical Systems and Signal Processing， 2001， 15（2）：287-302.

LIU H Y， XU Z N， XI J J. Wire finishing mill rolling bearing fault diagnosis based on feature extraction and BP neural network ［J］. Sensors & Transducers， 2014， 180（10）：190-197.

ZHOU S H， QIAN S L， CHANG W B， et al. A novel bearing multi-fault diagnosis approach based on weighted permutation entropy and an improved SVM ensemble classifier［J］.Sensors， 2018，18（6）： 1934

郑近德，姜战伟，代俊习，等. 基于VMD的自适应复合多尺度模糊熵及其在滚动轴承故障诊断中的应用［J］. 航空动力学报， 2017，32（7）：1683-1689.

姚德臣，杨建伟，程晓卿，等. 基于多尺度本征模态排列熵和SA-SVM的轴承故障诊断研究［J］. 机械工程学报， 2018， 54（9）：168-176.

HUANG G B， ZHU Q Y， SIEW C K. Extreme learning machineaheory and applications［J］. Neurocom-pulin， 2006（70）：489-501

SHAO H D， JIANG H K， LI X Q， et al. Intelligent fault diagnosis of rolling bearing using deep wavelet auto-encoder with extreme learning machine［J］. Knowledge-Based Systems， 2018，140：1-14.

RODRIGUEZ N， LAGOS C， CABRERA E， et al. Extreme learning machine based on stationary wavelet singular values for bearing failure diagnosis［J］. Studies in Informatics and Control， 2017， 26（3）：287-294.

HUANG G B， ZHOU H， DING X， et al. Extreme learning machine for regression and multiclass classification［J］. IEEE Transactions on Systems， Man， and Cybernetics-Part B： Cybernetics， 2012， 42（2）：513-529.

QIN B， SUN G D， ZHANG L Y， et al. Fault features extraction and identification based rolling bearing fault diagnosis［J］. Journal of Physics Conference Series， 2017， 842（1）：012055.

LI K， SU L， WU J J. A rolling bearing fault diagnosis method based on variational mode decomposition and an improved kernel extreme learning machine［J］. Applied Sciences， 2017， 7（10）：1004.

周绍磊，廖剑，史贤俊. 基于Fisher准则和最大熵原理的SVM核参数选择方法［J］. 控制与决策， 2014，29（11）：1991-1996.

张文兴，陈肖洁.基于核极化的特征选择在LSSVM的应用［J］. 计算机工程与应用， 2017， 53（19）：164-167.

王裴岩，蔡东风. 普适性核度量标准比较研究［J］.软件学报， 2015， 26（11）：2856-2868.

范永东. 模型选择中的交叉验证方法综述［D］. 太原：山西大学， 2013：15-18.

CRISTIANINI N， SHAWETAYLOR J， ELISSEEFF A， et al. On kernel target alignment［J］. Advances in Neural Information Processing Systems， 2001：367-373.

Case Western Reserve University. Bearing data ［EB/OL］. Cleveland：CWRU［2013-09-04］. http：//csegroups.case.edu/bearingdatacenter/pages/download-data-filehttp://csegroups.case.edu/bearingdatacenter/pages/download-data-file.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Application of EMPE and KP-KELM in Fault Diagnosis of Planetary Gearbox

Bearing Feature Extraction Method Based on the Time Subsequence

Bearing Fault Diagnosis Based on Parameter Optimized VMD and ELM with Improved SSA

Fault Diagnosis of Rolling Bearings Based on Two-step Transfer Learning and EfficientNetV2

A Fault Feature Extraction Method of Rolling Bearings Based on Optimized VMD and UMAP

Related Author

No data

Related Institution

State Key Laboratory of Smart Manufacturing for Special Vehicles and Transmission System， Baotou 014032， Chian

IOT Engineering College, Wuxi City College of Vocational Technology

School of Information Management, Wuxi Institute of Communications Technology

School of Mechatronics Engineering, Henan University of Science and Technology

SIPPR Engineering Group Co., Ltd.

⁰