铁道科学与工程学报2025，Vol.22Issue(3)：942-953,12.DOI:10.19713/j.cnki.43-1423/u.T20240761
基于RIME-VMD的高速列车横向减振器故障诊断
Fault diagnosis of transverse damper for high-speed trains based on RIME-VMD
摘要
Abstract
To solve the problem of more difficult feature extraction of variational modal decomposition(VMD)in the fault diagnosis of transverse damper of high-speed trains,a feature extraction method was proposed,according to the Frost and Ice Algorithm(RIME)optimizing the Variational Modal Decomposition(VMD)with the minimum envelope entropy as the fitness function.First,the optimal parameter combinations of the number of modal(IMF)components and the penalty factor of the VMD for different fault states were optimized using the Frost and Ice Algorithm(RIME).Secondly,the crag values and correlation coefficients of each IMFs component are computed.The first 4 orders of IMF components with larger crag values were selected respectively and the first 3 orders of IMFs with higher correlation coefficients were selected from among the 4 IMFs components with larger crag values.Finally,the multi-scale singular entropy,sample entropy,and alignment entropy were calculated as the fault feature values.These values were then combined with the t-distributed stochastic nearest neighbor embedding(t-SNE)algorithm to reduce dimensionality and remove redundant feature information,and then the reduced and fused feature matrices were inputted into the support vector machine(SVM)one by one,so as to realize the identification of the different faulty parts of the transverse damper in the high-speed train.The simulation results show that compared with the Gray Wolf Algorithm(GWO)optimized Variational Modal Decomposition(VMD)method,the RIME-VMD method utilizes the efficient search and development capabilities of the Frost and Ice Algorithm,which can more quickly find the globally optimal combinations of the number of decomposition layers and the penalty factor parameters in variational modal decomposition for high-speed trains under different working conditions.This can improve the robustness of the decomposed signals of the VMD.By using signal reconstruction methods,fault features can be effectively extracted,enabling efficient and accurate identification of transverse damper faults in high-speed trains.Although the original variational modal decomposition(VMD)method is faster,the manual trial-and-error cost of the original VMD parameters is higher,which cannot meet the requirements of high-speed train fault diagnosis.The results of the study can provide a reference for further optimization of transverse damper fault diagnosis and safe operation of high-speed trains.关键词
转向架/变分模态分解/霜冰算法/故障诊断/多尺度奇异熵Key words
bogie/variational modal decomposition/rime-ice algorithm/fault diagnosis/multi-scale singular entropy分类
交通运输引用本文复制引用
秦永峰,李刚,齐金平,王建帅..基于RIME-VMD的高速列车横向减振器故障诊断[J].铁道科学与工程学报,2025,22(3):942-953,12.基金项目
国家自然科学基金资助项目(71861021,72361019) （71861021,72361019）
甘肃省高等学校科研项目(2018A-026) （2018A-026）
