基于MEEMD香农熵-LSSVM的高速列车蛇行失稳诊断方法

发布时间：2018-01-28 00:43

本文关键词： 蛇行运动改进的集合经验模态分解(MEEMD) Hilbert-Huang变换(HHT) 香农熵最小二乘法支持向量机(LSSVM)　出处：《计算机应用研究》2017年04期 　论文类型：期刊论文

【摘要】：针对列车高速运行时易出现蛇行失稳这一问题,提出了一种改进的集合经验模态分解(modified ensemble empirical mode decomposition,MEEMD)香农熵—最小二乘法支持向量机(least squares support vector machine,LSSVM)的高速列车蛇行失稳诊断方法。首先通过MEEMD对列车330~350 km/h时转向架构架的横向振动信号进行分解,得到固有模态函数(intrinsic mode function,IMF),再通过Hilbert变换(HT)分析其时频聚集性,同时提取IMF分量的香农熵特征,最后用LSSVM进行训练和识别。结果表明,转向架蛇行失稳状态下的时频分布的聚集性较正常状态下好,并且MEEMD香农熵-LSSVM方法的识别率和计算耗时优于EEMD-SVM方法,识别率达到96.67%。
[Abstract]:In view of the problem that the train is prone to snakewalk instability when it is running at high speed. An improved set empirical mode decomposition (EMD) modified ensemble empirical mode decomposition is proposed. Shannon Entropy-least Squares support Vector Machine (squares support vector machine). First, the transverse vibration signal of the bogie frame is decomposed by MEEMD when the train is 330 ~ 350 km/h. The intrinsic mode function of the intrinsic mode function is obtained, and the time-frequency clustering is analyzed by Hilbert transform. At the same time, the Shannon entropy feature of IMF component is extracted and finally trained and recognized by LSSVM. The results show that the aggregation of time-frequency distribution is better than that under normal condition. The recognition rate and computation time of MEEMD Shannon Entropy-LSSVM method is better than that of EEMD-SVM method, and the recognition rate is 96.67.
【作者单位】：西南交通大学机械工程学院;
【基金】：国家自然科学基金资助项目(51475387) 中央高校基本业务费专项基金资助项目(2682014CX033) 四川省科技创新苗子工程资助项目(2015102)
【分类号】：U279.3;TP18
【正文快照】： 持向量机(LSSVM)0引言由于机车轮对具有一定锥度,当其沿着钢轨滚动并且速度达到某一临界值时,外界自激频率与列车系统固有频率接近,就会产生共振,此时轮对绕着轨道中心线一边横移一边摇头向前运动,即蛇行运动[1~3]。当列车发生蛇行运动时,如果不对列车速度加以控制,其振幅就会

【相似文献】