摘要
集合经验模态分解(EEMD,ensemble empirical mode decomposition)对信号进行分解,得到的模态函数(IMF,Intrinsic model function)在2端点存在严重的发散现象,如果将分解结果直接应用到故障诊断系统中,会导致诊断的准确率下降。首先将支持向量机(SVM,support vector machine)和EEMD算法结合进行信号分解,并利用仿真信号进行可靠性分析;其次对SVM(support rector machine)-EEMD分解的分量进行选择后再分解并构建能量向量,最后和卷积神经网络结合,构建滚动轴承故障诊断模型并通过实验验证。结果表明,改进EEMD算法可以有效缓解端点发散问题,构建的故障诊断模型提高了故障诊断精度。
EEMD(ensemble empirical mode decomposition)is an analysis method for signal decomposition.However,there is serious divergence in the two endpoints of its modal function(IMF).If the decomposition results are directly applied to the fault diagnosis system,the diagnosis accuracy will decrease.In the paper,support vector machine(SVM)and EEMD algorithm were combined to decompose signal and the reliability analysis was conducted with simulation signal.After selecting the components of SVM-EEMD decomposition,the signal was decomposed further and the energy vector was constructed.Finally,with a combination of SVM-EEMD and convolutional neural network,rolling bearing fault diagnosis model was constructed and verified by experiment.The experimental comparison results show that the improved EEMD algorithm can effectively solve the problem of the endpoints divergence,and the fault diagnosis model constructed improves the fault diagnosis accuracy.
作者
何江江
李孝全
赵玉伟
张保山
丁海斌
HE Jiangjiang;LI Xiaoquan;ZHAO Yuwei;ZHANG Baoshan;DING Haibin(Gradute School of Air Force Engineering University,College of Air Force Engineering University,Xi’an 710053P.R.China;Air and Missile Defense College of Air Force Engineering University,Xi’an 710053P.R.China;Unit 92095,Taizhou 318050,Zhejiang,P.R.China;Training Base of Army Engineering University of PLA,Xuzhou 21004,Jiangsu,P.R.China)
出处
《重庆大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2020年第1期82-89,共8页
Journal of Chongqing University
基金
国家自然科学基金资助项目(51405505).
