基于EEMD和球SVM的高速列车转向架故障估计  

Running State Estimation of High-speed Train Based on EEMD and Hyper-sphere Support Vector Machines

作　　者：翟冰[1] 金炜东[1] 秦娜[1] 

机构地区：[1]西南交通大学电气工程学院,成都610031

出　　处：《计算机测量与控制》2014年第8期2356-2359,共4页Computer Measurement &Control

基　　金：国家自然科学基金重点项目(61134002)

摘　　要：基于监测数据评估高速列车空气弹簧、抗蛇行减振器和横向减振器关键部件的运行状态,针对故障状态下车体构架横向加速度的非平稳信号,提出IMF能量矩与改进的多类超球支持向量机相结合对车体运行状态估计,改进的的超球支持向量机对球外与球内重叠区域的样本用不同的决策,具有更好的分类效果;实验数据仿真分析表明,在速度变化下列车故障识别率稳定在87%以上,证明所用方法能够提取到故障状态下的典型特征,改进的支持向量分类器并能很好的估计出高速列车的故障状态。Based on monitoring data, the running state of key components about air springs , resist sinusoidal vibration absorber and transverse shock absorbers damper of high--speed train were estimated. Aiming at Lateral acceleration for high--speed train bogie in fault, and a typical non--stationary complex vibration signal, a running state estimation method of combining IMF energy moment and improved Hyper--sphere support vector machines methods was proposed. The improved Hyper--spbere support vector machines methods based on improved samples in different parts with different classification policy. Experimental results show that under different speeds, the high speed train recognition rate of damper fault are steady at 87% above. It proves that the improved method can be used to extract the characteristics of the fault condition, and can estimate running state of the high--speed train effectively.

关 键 词：转向架 阈值消噪 EEMDIMF能量矩 超球支持向量机 

分 类 号：U283.2[交通运输工程—交通信息工程及控制]