CEEMD-VMD与参数优化SVM结合的托辊轴承故障诊断  

Roller Bearing Fault Diagnosis Combined CEEMD-VMD and Parameter Optimization SVM

作　　者：贺志军 李军霞 刘少伟 秦志祥 HE Zhijun;LI Junxia;LIU Shaowei;QIN Zhixiang(College of Mechanical and Vehicle Engineering,Taiyuan University of Technology,Taiyuan 030024,China;Shanxi Research Center of Mining Fluid Control Engineering,Taiyuan 030024,China;National-local Joint Laboratory of Mining Fluid Control Engineering,Taiyuan 030024,China)

机构地区：[1]太原理工大学机械与运载工程学院,太原030024 [2]山西省矿山流体控制工程技术研究中心,太原030024 [3]矿山流体控制国家地方联合工程实验室,太原030024

出　　处：《机械科学与技术》2024年第3期402-408,共7页Mechanical Science and Technology for Aerospace Engineering

基　　金：国家自然基金面上项目(52174147);中央引导地方科技发展资金项目(YDZJSX2021A023);晋中市科技重点研发项目(Y211017)。

摘　　要：针对托辊轴承工作环境复杂、提取故障特征困难等问题,提出一种基于互补集合经验模态分解(Complementary ensemble empirical mode decomposition, CEEMD)和变分模态分解(Variational modal decomposition, VMD)相结合的降噪方法。首先,利用CEEMD将采集到的信号进行分解,依据相关系数和峭度筛选分量并进行重构,生成新的信号;然后,利用VMD将新的信号进行再分解,并基于包络熵和包络谱峭度组合的复合指标优选本征模态分量(Intrinsic mode functions, IMF);最后,提取相应的特征输入樽海鞘群优化支持向量机(Salp swarm optimization support vector machine, SSO-SVM)模型完成故障诊断。实验结果表明:对于正常轴承、轴承内圈故障、轴承外圈故障三种情况,诊断准确率达97.78%。与单一降噪方法相比,该方法可以有效提高故障信号的信噪比,降噪效果明显。In order to solve the difficulty in extracting fault features of roller bearings under complex working environment,a noise reduction method was proposed based on the combination of complementary ensemble empirical mode decomposition(CEEMD)and variational modal decomposition(VMD).Firstly,the collected signals are decomposed by CEEMD,and the components are screened and reconstructed according to the correlation coefficient and kurtosis to generate new signals.Then,VMD was used to decompose the new signal,and the intrinsic mode functions(IMF)were optimized based on the composite index of the combination of envelope entropy and envelope spectrum kurtosis.Finally,the corresponding features were extracted and input into salp swarm optimized support vector machine(SSO-SVM)model to complete the fault diagnosis.The experimental results show that the diagnosis accuracy of normal bearing,bearing inner ring fault and bearing outer ring fault is up to 97.78%.Compared with the single noise reduction method,this method can effectively improve the signal noise ratio of fault signal,and the noise reduction effect is obvious.

关 键 词：变分模态分解 托辊轴承 樽海鞘群算法 支持向量机 故障诊断 

分 类 号：TH113.1[机械工程—机械设计及理论] TH133.33