基于IMDE与PSO-SVM组合算法的电机故障诊断研究  

Research on Motor Fault Signal Diagnosis based on Combination Algorithm of IMDE with PSO-SVM

作　　者：孙长胜[1] 刘亚楠[2] 汪靖博 Sun Changsheng;Liu Yanan;Wang Jingbo(School of Architectural Engineering,Xuchang Vocational and Technical College,Xuchang 461000,China;School of Mechanical and Automotive Engineering,Xuchang Vocational and Technical College,Xuchang 461000,China;Luoyang Yusan Construction Inspection Co.,Ltd.,Luoyang 471000,China)

机构地区：[1]许昌职业技术学院建筑工程学院,河南许昌461000 [2]许昌职业技术学院机械与汽车工程学院,河南许昌461000 [3]洛阳市豫三建筑检测有限公司,河南洛阳471000

出　　处：《防爆电机》2024年第5期29-31,41,共4页Explosion-proof Electric Machine

摘　　要：为了提高电机故障信号的特征提取能力,设计了一种基于改进多尺度散布熵(IMDE)与粒子群算法-支持向量机(PSO-SVM)组合算法的振动信号特征来实现电机故障诊断的方法。研究结果表明:熵值曲线达到了更平滑程度并且可以满足收敛要求,此外对于混叠区而言也比MDE达到了更优的区分度,实现了鲁棒性的显著提升。采用未降噪初始信号进行分类时准确率只达到86.81%,说明EEMD分解性能比EMD更优。本设计方法满足可靠性与优越性要求,可适用于其它机械传动设备故障诊断领域。In order to improve the feature extraction ability of motor fault signals,a method based on the improved multi-scale dispersal entropy(IMDE)and the particle swarm optimization-support vector machine(PSO-SVM)combination algorithm is designed to realize the motor fault diagnosis by vibration single.The research results show that the entropy curve has achieved a much smoother level and can meet the convergence requirements.In addition,it has also achieved better differentiation than that of MDE for the aliasing region,which has significantly improved the robustness.When the initial signal without noise reduction is used for classification,the accuracy is only 86.81%.It indicates that the decomposition performance of EEMD is better than that of EMD.This design method meets the requirements of reliability and superiority,which can be applied to fault diagnosis fields of other mechanical transmission equipment.

关 键 词：故障诊断 电机 改进多尺度散布熵 鲁棒性 

分 类 号：TM301.3[电气工程—电机]