基于Hilbert包络谱奇异值和IPSO-SVM的滚动轴承故障诊断研究

Study on the Rolling Bearing Fault Diagnosis based on the Hilbert Envelope Spectrum Singular Value and IPSO-SVM

针对表征滚动轴承故障信号特征难提取及支持向量机结构参数依据经验选取,致使故障分类模型的精度、泛化能力差的问题,提出一种基于Hilbert包络谱奇异值和改进粒子群(Improved particle swarm optimization,IPSO)优化支持向量机(Support vector machine,SVM)的滚动轴承状态辨识方法。首先,利用经验模态分解(Empirical mode decomposition,EMD)所采集的滚动轴承信号,并将所获相关程度较大的本征模式分量(Intrinsic mode function,IMF)进行Hilbert解调包络分析来获取包络矩阵,并在此基础上进行奇异值分解。其次,利用IPSO算法优化SVM的惩罚系数和高斯核系数两个结构参数,据此建立滚动轴承故障分类模型;并利用美国凯斯西储大学轴承数据验证了方法的有效性。实验结果表明:与基于BP、SVM的故障分类模型相比,Hilbert包络谱奇异值和IPSO优化SVM的滚动轴承故障诊断分类模型具有更高的精度、更强的泛化能力。

For the problem that the characterization of the gear fault signal feature is difficult to extract and the structure parameters selection of support vector machine（ SVM） are based on experience leads the poor precision and generalization ability of fault state recognition,a method that IPSO- SVM rolling bearing fault diagnosis based on the Hilbert envelope spectrum singular value is proposed. Firstly,the rolling bearing signal is divided by EMD,it selects IMFs that contains main characteristics of signal for Hilbert demodulation envelope analysis to obtain envelope matrix and the singular value decomposition is carried out. Secondly,the IPSO algorithm is used to optimize the penalty coefficient and the structural parameters of SVM to set up the rolling bearing fault classification model. And by using the bearing data of Case Western Reserve University,the validity of the method is verified. The experimental results show that IPSO- SVM rolling bearing fault diagnosis based on the Hilbert envelope spectrum singular value compared with the fault classification model based on BP,SVM has higher precision and stronger generalization ability.