Scholarlink

<p>The normal operation of the bogie is essential for ensuring the safety and stability of high-speed trains. Due to the insufficiency and non-stationarity of operational data from key bogie components, this study proposes a state prediction method that integrates a diffusion model, variational autoencoder (VAE), and attention mechanism. Firstly, the DM-VAE data augmentation method is employed, leveraging the generative capability of the VAE and the data enhancement strength of the diffusion model to mitigate data scarcity. Secondly, an improved Transformer algorithm is designed to adaptively capture non-stationary features, improving prediction accuracy. Finally, the proposed method is validated through a case study on the bogie's air spring. The results demonstrate that the proposed approach achieves improved predictive performance and accuracy.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A method for predicting key bogie component status under limited and non-stationary data conditions

  • 期刊论文

  • 发表日期：

    2025年8月8日

查看全文
  • Kai Zhang,
  • Zhe Wei,
  • Lei Wang,
  • Duo Xu,
  • Guotian Huang

摘要

The normal operation of the bogie is essential for ensuring the safety and stability of high-speed trains. Due to the insufficiency and non-stationarity of operational data from key bogie components, this study proposes a state prediction method that integrates a diffusion model, variational autoencoder (VAE), and attention mechanism. Firstly, the DM-VAE data augmentation method is employed, leveraging the generative capability of the VAE and the data enhancement strength of the diffusion model to mitigate data scarcity. Secondly, an improved Transformer algorithm is designed to adaptively capture non-stationary features, improving prediction accuracy. Finally, the proposed method is validated through a case study on the bogie's air spring. The results demonstrate that the proposed approach achieves improved predictive performance and accuracy.