基于EEMD与GWO-SVM的石化机组轴承故障诊断

Fault Diagnosis of Petrochemical Unit Bearings Based on EEMD and GWO-SVM

针对石化机组轴承做故障分类时,传统支持向量机的分类性能受自身参数选择的影响识别准确率不高的问题,提出一种基于集合经验模态分解和改进支持向量机的石化机组轴承故障诊断。首先利用集合经验模态分解(ensemble empirical mode decomposition,EEMD)与样本熵(sample entropy,SE)对原始信号进行特征提取,采用灰狼算法优化支持向量机(gray wolf optimization algorithm support vector machine,GWO-SVM)的方法得到最优参数,构建石化机组轴承故障诊断模型。最后以实际石化机组数据集进行诊断分析,并通过与未优化的支持向量机和传统优化算法的支持向量机进行对比,表明该文所提方法的有效性和优越性。

To address the problem that the classification performance of traditional support vector machine is affected by the selection of its own parameters when classifying faults of petrochemical unit bearings,a fault diagnosis of petrochemical unit bearings based on ensemble empirical modal decomposition and improved support vector machine is proposed.Firstly,use ensemble empirical mode decomposition(EEMD)and sample entropy(SE)to extract features from the original signal,and use gray wolf optimization algorithm support vector machine(GWO-SVM)to obtain the optimal parameters to construct a petrochemical unit bearing fault diagnosis model.Finally,the diagnosis analysis is performed with the actual petrochemical unit data set,and the effectiveness and superiority of the proposed method is demonstrated by comparing it with the unoptimized support vector machine and the support vector machine of the traditional optimization algorithm.