摘要
针对轨道交通车辆轴箱振动信号的非平稳性特征,提出了一种基于经验模态分解(EMD)与BP神经网络的车轮多边形阶数诊断、识别方法。该方法首先对轴箱振动信号进行EMD分解,信号分解为若干个IMF分量之和,再选取若干个包含主要能量信息的IMF分量,提取各分量的能量与峭度系数构造特征向量,然后将特征向量作为神经网络的输入,最后对神经网络进行训练来识别车轮多边形阶数。在试验台开展了某型号动车组不同速度等级工况下车轮多边形故障模拟试验,设置了17阶和24阶两种车轮多边形工况。试验结果表明,基于EMD和神经网络的方法能够准确有效地识别出车轮多边形阶数。基于试验台模拟试验建立了车轮多边形阶数在线识别数据库,能够识别、诊断车轮磨耗信号,提前预判车轮状态,辅助检测与维护决策,可为轨道交通车辆的安全运营与维护提供方法参考和理论支撑。
To address the issue of non-stationarity in axle box vibration signals of rail transit vehicles,a diagnosis and identification method for wheel polygon order based on empirical mode decomposition(EMD)and BP neural network is proposed.Firstly,the EMD decomposition of the axle box vibration signal is carried out,and the signal is decomposed into the sum of several IMF components,then several IMF components containing main energy information are selected,the energy and kurtosis coefficient of each component are extracted,and the eigenvector is constructed,and then the eigenvector is used as the input of the neural network,and finally the neural network is trained to identify the wheel polygon order.The fault simulation test of wheel polygon under different speeds of a certain type of EMU was carried out on the test bench,and two wheel polygon conditions of 17th order and 24th order were set.The test results indicate that the method based on EMD and neural networks can accurately and effectively identify the wheel polygon order.Based on the simulation test on the test bench,an online identification database of wheel polygon order is established,which can identify and diagnose the wheel wear signal,predict the wheel state in advance,assist in detection and maintenance decision-making,and provide method reference and theoretical support for the safe operation and maintenance of rail transit vehicles.
作者
闫中奎
李春超
马丽英
贺竹林
YAN Zhongkui;LI Chunchao;MA Liying;HE Zhulin(CRRC Qingdao Sifang Locomotive&Rolling Stock Co.,Ltd.,Qingdao 266111,China)
出处
《铁道车辆》
2025年第3期71-76,共6页
Rolling Stock
作者简介
第一作者:闫中奎(1990-),男,硕士,工程师。E-mail:yanzhongkui@cqsf.com。
