基于MEEMD和GA-SVM的列车车轮多边形故障识别方法.PDF

第38 卷 第3 期 噪 声 与 振 动 控 制 Vol 38 No.3 2018 年6 月 NOISE AND VIBRATION CONTROL Jun. 2018 文章编号：1006-1355(2018)03-0157-05+ 197 基于MEEMD 和GA-SVM 的列车车轮多边形故障 识别方法 陈 博，陈光雄 （西南交通大学 机械工程学院，成都 610031 ） 摘 要：根据列车车轮振动信号的非平稳特性，提出一种基于改进的集合经验模态分解（MEEMD ）和遗传算法支持 向量机（GA-SVM）的诊断方法，用于识别车轮多边形故障。该方法对车轮轴箱垂向振动信号进行MEEMD 分解，依据 各固有模态函数（IMF）分量的峭度值和能量值选取出主要IMF 分量。利用希尔伯特变换求取主要IMF 分量的包络谱， 并计算包络谱熵。将包络谱熵值归一化后作为特征向量输入GA-SVM 中进行训练和识别。对实测信号进行分析的结 果表明该方法能有效识别出车轮多边形故障，识别准确率可达到95 % 。 关键词：振动与波；车轮多边形识别；改进的集合经验模态分解；遗传算法支持向量机；包络谱熵 中图分类号：TP206.3 ；U260.331.1 文献标志码：A DOI 编码：10.3969/j.issn. 1006-1355.2018.03.030 Fault Diagnosis Method of Wheel Polygonization of Trains based on MEEMD and GA-SVM CHEN Bo , CHEN Guangxiong （School of Mechanical Engineering, Southwest Jiaotong University, Chengdu 610031, China ） Abstract : According to the non-steady characteristics of train wheel vibration signals, a diagnostic method based on modified ensemble empirical mode decomposition (MEEMD) and genetic algorithm support vector machine (GA-SVM) is proposed for identifying wheel polygonization faults. First of all, the vertical vibration signals of the wheel axle-box are decomposed into a number of intrinsic mode function (IMF) components by MEEMD. According to the kurtosis and energy values of the components, the main IMF components are selected. Then, Hilbert transform is used to obtain the envelope spectra of the main IMF components. And the envelope spectrum entropy values are calculated by the envelope spectrum. Finally, the normalized envelope spectrum entropy values are employed as the