SO-VMD和IHFDE在旋转机械耦合故障辨识中的应用  

Application of SO-VMD and IHFDE in couplingfault identification of rotating machinery

作　　者：张文军 宋琳璐 左小勇[2] 王冠华 ZHANG Wenjun;SONG Linlu;ZUO Xiaoyong;WANG Guanhua(College of Mechanical and Electrical Engineering,Nanyang Vocational College of Agriculture,Nanyang 473000,China;School of Automotive Engineering,Xiangyang Auto Vocational Technical College,Xiangyang 441021,China;College of Mechanical and Vehicle Engineering,Hunan University,Changsha 410082,China)

机构地区：[1]南阳农业职业学院机电工程学院,河南南阳473000 [2]襄阳汽车职业技术学院汽车工程学院,湖北襄阳441021 [3]湖南大学机械与运载工程学院,湖南长沙410082

出　　处：《机电工程》2025年第4期714-725,共12页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金资助项目(52005170)。

摘　　要：采用传统旋转机械故障诊断模型诊断单点故障而忽略多点故障缺陷,无法准确判断旋转机械的故障来源,提出了一种基于蛇优化器的优化变分模态分解(SO-VMD)、改进层次波动散布熵(IHFDE)和支持向量机(SVM)的旋转机械耦合故障诊断方法。首先,以模态分量的最大互信息系数为适应度函数,采用蛇优化器对变分模态分解的参数进行了优化,并对旋转机械振动信号进行了分解以得到模态分量;然后,对各模态分量的IHFDE特征值进行了提取,从而构建了故障特征矩阵;最后,将故障特征输入至SVM分类器中进行了分类识别,并实现了对旋转机械的故障诊断。利用滚动轴承和齿轮箱的多点故障数据集进行了实验分析,从信号处理和特征提取两方面进行了对比分析。研究结果表明:SO-VMD-IHFDE故障诊断方法在诊断旋转机械的单点和多点故障时分别取得了98.75%和100%的识别精度,验证了该方法的有效性。SO-VMD方法能够有效去除信号中的干扰噪声,提高特征的质量。和未采用SO-VMD方法得到的诊断结果相比,滚动轴承和齿轮箱的诊断准确率分别提高了3.33%和5.42%。IHFDE方法能够有效反映旋转机械的故障特性,准确率高于其他广泛使用的特征提取方法。旋转机械的故障诊断结果验证了改进层次分析在诊断准确率方面要优于粗粒化处理和传统层次分析。The traditional fault diagnosis model of rotating machinery only pays attention to the single point fault and neglects the simultaneous occurrence of multiple points of fault,in order to solve the problem,a rotating machinery coupling fault diagnosis method was proposed based on the snake optimizer optimized variational mode decomposition(SO-VMD),improved hierarchical fluctuation dispersion entropy(IHFDE)and support vector machine(SVM).Firstly,using the maximum mutual information coefficient of the modal components as the fitness function,the snake optimizer was used to optimize the parameters of variational modal decomposition,and the vibration signal of rotating machinery was decomposed to obtain modal components.Then,the IHFDE eigenvalues of each modal component were extracted to construct a fault feature matrix.Finally,the fault features were input into a support vector machine classifier for classification and recognition,achieving fault diagnosis of rotating machinery.The experimental analysis was conducted using rotating machinery fault datasets,and comparisons were made from two aspects:signal processing and feature extraction.The research results show that the SO-VMD-IHFDE fault diagnosis method achieves 98.75%and 100%accuracy in diagnosing single point and multi-point faults of rotating machinery,and it verifies its validity.SO-VMD method can effectively remove interference noise in the signal,improve the quality of features,and respectively improve the diagnosis accuracy of rolling bearings and gearboxes by 3.33%and 5.42%,compared to the diagnostic results obtained without using SO-VMD method.The IHFDE method can effectively reflect the fault characteristics of rotating machinery,and the accuracy is higher than other widely used feature extraction methods.The fault diagnosis results of rotating machinery verify that the improved hierarchical analysis is superior to coarse-grained processing and traditional hierarchical analysis in terms of diagnostic accuracy.

关 键 词：旋转机械 耦合故障诊断 变分模态分解 改进层次波动散布熵 蛇优化器 多点故障 耦合故障 信号高频特征信息 

分 类 号：TH133[机械工程—机械制造及自动化]