摘要
为提高利用液压泵振动信号进行故障诊断的准确率和减小诊断时间,采用集合经验模态分解(EEMD)的方式来提取振动信号特征,并将其作为液压泵故障诊断的数据集。在此基础上利用支持向量机(SVM)与深度神经网络(DNN)进行故障诊断,最后通过验证数据集检验模型诊断故障的准确程度。结果表明:EEMD-SVM在液压泵故障诊断方面具有较好的性能,与神经网络故障诊断模型相比,支持向量机模型在液压泵故障诊断方面具有更高的准确率和更短的诊断时间。
In order to improve the accuracy and reduce the diagnosis time of hydraulic pump fault diagnosis by using vibration signal, the ensemble empirical mode decomposition(EEMD) method is used to extract vibration signal characteristics, and it is used as the data set of hydraulic pump fault diagnosis. On this basis, a support vector machine(SVM) and deep neural network(DNN) are used for fault diagnosis. Finally, the accuracy of the model fault diagnosis is verified by validating data sets. The results show that EEMD-SVM has better performance in fault diagnosis of hydraulic pumps. Compared with the neural network fault diagnosis model, the support vector machine model has higher accuracy and shorter diagnosis time in fault diagnosis of hydraulic pumps.
作者
袁兵
余佳翰
邹永向
Yuan Bing;Yu Jiahan;Zou Yongxiang
出处
《起重运输机械》
2019年第20期90-95,共6页
Hoisting and Conveying Machinery
关键词
液压泵
集合经验模态分解
支持向量机
故障诊断
hydraulic pump
collective empirical mode decomposition
support vector machine
fault diagnosis
