基于深度核极限学习机的重载机车车钩摆角识别

打开文本图片集

中图分类号：U270.1 文献标志码：B doi：10.20213/j.cnki.tdcl.2025.01.018

文章编号：1002-7602（2025）01-0152-07

Abstract： In view of the shortcomings of traditional coupler swing angle monitoring methods，such as high monitoring cost and poor long-term monitoring stability，this paper proposes an identification method based on deep kernel extreme learning machine for heavy-duty locomotives to realize the accurate identification of coupler swing angle through carbody vibration acceleration.Firstly，the dynamic model of a heavy-hual locomotive is built，the three-way vibration aceleration signal of carbody is extracted by dynamic simulation，the signal entropy value is calculated，and the sensitive channel for coupler swing angle identification is screened out by correlation analysis.Furthermore，the multidimensional signal characteristics of the coupler swing angle identification sensitive channel are extracted；Kernel principal component analysis （KPCA） is used to fuse and reduce the dimension of multidimensional features of multichannel signals，to reduce the feature dimension and achieve theeffctof noisereduction.Finall，thedeep kernelextreme learning machine（DKELM） isusedtobuilda nonlinear mapping modelof carbody lateral acceleration and coupler swing angle.The simulation results show that the coupler swingangle identification method proposed inthis paper has theadvantagesof high eficiencyandlow monitoring cost and can provide a theoretical basis for the coupler swing angle monitoring of heavy-duty locomotives.

Key words： heavy haul locomotive;Kernel Extreme Learning Machine;dynamic monitoring；coupler yav angle;real-time identification

随着重载铁路运输的不断发展，重载列车的载重和编组长度不断提高。（剩余9469字）