基于分布式SVM的高速列车运行状态评估  

High-Speed Train Running Condition Assessment Based on Distributed SVM Algorithm

作　　者：熊定鸿[1] 

机构地区：[1]西南交通大学电气工程学院,成都610031

出　　处：《现代计算机（中旬刊）》2016年第9期38-42,共5页Modern Computer

基　　金：国家自然科学基金(No.61134002)

摘　　要：随着大规模数据信息不断涌现,支持向量机作为一种经典高效的监督式的机器学习方法也在不断地进步与创新,实现分布式支持向量机算法。与此同时,高速列车的快速发展,高速列车安全性问题逐渐引起人们的重视。将分布式SVM算法与Hadoop平台相结合,采用EEMD算法提取IMFs特征,运用分布式SVM算法对高速列车振动数据进行深层特征提取并进行故障分类。实验表明通过对较好通道的统计结果来看,列车故障辨识的识别率为96%,故障定位识别率为89%,且算法效率有一定的提升。With the production of large-scale data, the support vector machine （SVM） also has been improved and innovated continuously, as a tra- ditional and effective supervised method of machine learning, and developed the distribution SVM algorithm. At the same time, with the rapid development of high-speed trains, high-speed train security issues gradually attracted people＇s attention. Combines distributed SVM algorithm and Hadoop platform, uses distributed SVM algorithm for high-speed train vibration data to extract deep fault feature and clas- sify these failures. Experimental results show that, by the statistical results of the better channel, the train fault identification recognition rate of 96%, and the fault location identification rate of 89%.

关 键 词：高速列车监测数据 故障诊断 支持向量机 分布式计算 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]