摘要
针对列车轮对振动信号易受轮轨噪声影响、故障特征提取困难等问题,提出一种基于优化变分模态分解(Variational mode decomposition,VMD)和多尺度样本熵-能量(Multiscale sample entropy-energy,MSEEN)指标的故障诊断方法。首先搭建考虑轮轨接触关系的轮对振动实验台,分别进行正常、车轮扁疤、车轴裂纹及扁疤-裂纹耦合故障状态下的轮对振动测试。其次,利用遗传算法,以样本熵、相关系数和均方误差为适应值搜索VMD的最佳分解个数及分解中心频率。然后基于优化VMD分解不同状态下的轮对振动信号并提取本征模态函数(Intrinsic mode function,IMF)分量的MSEEN指标。最后将指标与BP神经网络结合进行轮对故障诊断,总识别率达到94.44%。该方法可为实际运行工况中的列车轮对故障诊断提供借鉴。
In order to solve the problems that the vibration signal of the train wheelset is easily affected by the wheelrail noise and it is difficult to extract the fault characteristics,a fault diagnosis method based on optimized variational modal decomposition(VMD)and multiscale sample entropy-energy(MSEEN)indicators were proposed.Firstly,a wheelset vibration test bench considering the wheel-rail contact was built,and the wheelset vibration tests under the conditions of healthy,wheel flat,axle crack,and wheel flat-crack coupling fault were carried out respectively.Secondly,using genetic algorithm,the best decomposition number and decomposition center frequency of VMD were searched with the minimum sample entropy,correlation coefficient,and mean square error as fitness values.Then,based on the optimized VMD algorithm,the wheelset vibration signals in different states were decomposed and the MSEEN indicators of the intrinsic mode function(IMF)components were extracted.Finally,the indicators were combined with BP neural network to perform wheelset fault diagnosis,and the total recognition rate was found to reach 94.44%.The method in this paper may provide a basis for fault diagnosis of train wheelset in actual operating conditions.
作者
蒋宇涵
华春蓉
董大伟
熊丽波
王瑞
JIANG Yuhan;HUA Chunrong;DONG Dawei;XIONG Libo;WANG Rui(School of Mechanical Engineering,Southwest Jiaotong University,Chengdu 610031,China)
出处
《噪声与振动控制》
CSCD
北大核心
2021年第6期71-76,共6页
Noise and Vibration Control
基金
国家自然科学基金资助项目(51875482)。
