Scholarlink

<p>Traditional<?tk 4?> diagnostic methods for multifunction vehicle bus (MVB) faults often depend on feature extraction and classification, which typically require substantial expert experience and frequently yield low accuracy. To address this limitation, this paper introduces a MVB fault diagnosis method utilizing convolutional neural networks (CNNs). Initially, the method employs short-time Fourier transform (STFT) to convert the original vibration signals of MVB under various fault conditions into time-frequency images. Subsequently, a specific MVB fault diagnosis model, termed STCNN, is developed to conduct deep spatial feature learning on these two-dimensional signals. Fault classification is then achieved through a Softmax classifier. The model was tested on a MVB network dataset collected under diverse operating conditions on a test bench. The results demonstrated a fault detection accuracy of 99.68<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="41598_2025_89793_Article_IEq1.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="15" /> </InlineMediaObject> <EquationSource Format="TEX">\(\%\)</EquationSource> </InlineEquation>, significantly surpassing other methods and highlighting its superior performance in fault diagnosis.<?tk 0?></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

MVB fault diagnosis based on time-frequency analysis and convolutional neural networks

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年2月12日

查看全文
  • Xudong Song,
  • Zhibo Li,
  • Yang Liu

摘要

Traditional diagnostic methods for multifunction vehicle bus (MVB) faults often depend on feature extraction and classification, which typically require substantial expert experience and frequently yield low accuracy. To address this limitation, this paper introduces a MVB fault diagnosis method utilizing convolutional neural networks (CNNs). Initially, the method employs short-time Fourier transform (STFT) to convert the original vibration signals of MVB under various fault conditions into time-frequency images. Subsequently, a specific MVB fault diagnosis model, termed STCNN, is developed to conduct deep spatial feature learning on these two-dimensional signals. Fault classification is then achieved through a Softmax classifier. The model was tested on a MVB network dataset collected under diverse operating conditions on a test bench. The results demonstrated a fault detection accuracy of 99.68 \(\%\) , significantly surpassing other methods and highlighting its superior performance in fault diagnosis.