摘要
针对电机滚动轴承故障特征难以提取从而导致诊断准确率低的问题,提出了一种基于变分模态分解(Variational Modal Decomposition,VMD)结合小波包信息熵(Wavelet Packet Information Entropy,WPIE)的特征提取方法,并采用金豺优化(Golden Jackal Optimization,GJO)算法优化后的支持向量机(Support Vector Machine,SVM)进行电机滚动轴承的故障诊断。首先,利用VMD将采集到的信号进行分解,依据局部极小包络熵筛选出最优本征模态(Intrinsic Mode Function,IMF)分量;其次,利用小波包将最优IMF分量再分解,并提取信息熵作为特征向量矩阵;最后,采用GJO算法对支持向量机中的惩罚参数和核参数进行寻优选择,建立GJO-SVM故障诊断模型,将特征向量矩阵输入金豺算法优化支持向量机(GJO-SVM)故障诊断模型中进行故障诊断。将VMD结合小波包信息熵特征提取与VMD结合近似熵特征提取进行对比试验,试验结果表明,VMD结合小波包信息熵特征提取精度提高了2.5%,其特征提取更加优越;将金豺算法优化支持向量机(GJO-SVM)与粒子群优化(Porticle Swarm OPtimization,PSO)算法支持向量机(PSO-SVM)、果蝇优化算法(Fruit fly Optimation Algorithm,FOA)支持向量机(FOA-SVM)进行对比试验,试验结果表明,GJO-SVM其平均准确率达到99.16%,较PSO-SVM、FOA-SVM分别提高了2.5%、3.61%。金豺算法优化支持向量机(GJO-SVM)可以更加有效提取并诊断滚动轴承故障。
To address the problem of low diagnostic accuracy due to the difficulty in extracting fault features of rolling bearings in electric motors,a feature extraction method based on Variational Modal Decomposition(VMD)combined with Wavelet Packet Information Entropy(WPIE)is proposed.A Support Vector Machine(SVM)optimized by Golden Jackal Optimization(GJO)is used for the fault diagnosis of motor bearings.Firstly,the collected signal is decomposed by VMD and the optimal eigenmode component Intrinsic Mode Function(IMF)is filtered based on the local minimal envelope entropy;secondly,the wavelet packet is decomposed again and the information entropy is extracted as the feature vector matrix;finally,the penalty and kernel parameters in the support vector machine are optimally selected by the GJO algorithm,and the GJO-SVM fault diagnosis model is established,the feature vector matrix is input into Golden Jackal Optimizes algorithm the Support Vector Machine(GJO-SVM)for fault diagnosis.The VMD combined with wavelet packet information entropy feature extraction is compared with VMD combined with approximate entropy feature extraction,and the experimental results show that the accuracy of VMD combined with wavelet information entropy feature extraction is improved by 2.5%,and its feature extraction is more superior.The experimental results show that the average accuracy of GJO-SVM reaches 99.16%,which is 2.5%and 3.61%higher than that of PSO-SVM and FOA-SVM respectively.GJO-SVM can extract and diagnose bearing faults more effectively.
作者
纪京生
周莉
马向阳
JI Jingsheng;ZHOU Li;MA Xiangyang(School of Electrical and Information Engineering,Anhui University of Science and Technology,Huainan 232001,China)
出处
《现代制造工程》
CSCD
北大核心
2024年第2期128-136,共9页
Modern Manufacturing Engineering
关键词
变分模态分解
小波包信息熵
金豺优化算法
支持向量机
轴承故障诊断
Variational Modal Decomposition(VMD)
Wavelet Packet Information Entropy(WPIE)
Golden Jackal Optimization(GJO)algorithm
Support Vector Machine(SVM)
bearing fault diagnosis
