<p>Bearing fault diagnosis is crucial for the safe and stable operation of mechanical equipment. However, bearing signals are highly susceptible to noise interference, which complicates feature extraction. Existing multi-source data diagnostic methods still face challenges in effectively integrating signals and suppressing noise. To address these challenges, this paper proposes a multi-sensor data fusion and multi-scale quadratic convolutional neural network for intelligent bearing fault diagnosis. First, the method inputs vibration signals collected by multiple sensors into a time domain filter consisting of a quadratic convolutional network and a frequency domain filter based on a fully connected neural network for processing. The filtered signal is then passed into a multi-scale quadratic convolutional neural network, which utilizes quadratic neurons with strong feature extraction capabilities for bearing vibration signals. The extracted multi-scale features are further refined through a cross attention mechanism to capture more useful information, which is then classified. Experimental results conducted on the bearing datasets from Case Western Reserve University and Politecnico di Torino demonstrate that the proposed method outperforms other comparative models in noisy environments. At a signal-to-noise ratio of <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11071_2025_10918_Article_IEq1.gif" Format="GIF" Height="14" Rendition="HTML" Resolution="72" Type="Linedraw" Width="32" /> </InlineMediaObject> <EquationSource Format="TEX">\(-10\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mo>-</mo> <mn>10</mn> </mrow> </math></EquationSource> </InlineEquation>, the method achieves accuracies of 97.20% and 98.81%, respectively, verifying its excellent performance under complex noise interference conditions.</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

Multi-scale quadratic convolutional neural network for bearing fault diagnosis based on multi-sensor data fusion

  • Yingying Ji,
  • Jun Gao,
  • Xing Shao,
  • Cuixiang Wang

摘要

Bearing fault diagnosis is crucial for the safe and stable operation of mechanical equipment. However, bearing signals are highly susceptible to noise interference, which complicates feature extraction. Existing multi-source data diagnostic methods still face challenges in effectively integrating signals and suppressing noise. To address these challenges, this paper proposes a multi-sensor data fusion and multi-scale quadratic convolutional neural network for intelligent bearing fault diagnosis. First, the method inputs vibration signals collected by multiple sensors into a time domain filter consisting of a quadratic convolutional network and a frequency domain filter based on a fully connected neural network for processing. The filtered signal is then passed into a multi-scale quadratic convolutional neural network, which utilizes quadratic neurons with strong feature extraction capabilities for bearing vibration signals. The extracted multi-scale features are further refined through a cross attention mechanism to capture more useful information, which is then classified. Experimental results conducted on the bearing datasets from Case Western Reserve University and Politecnico di Torino demonstrate that the proposed method outperforms other comparative models in noisy environments. At a signal-to-noise ratio of \(-10\) - 10 , the method achieves accuracies of 97.20% and 98.81%, respectively, verifying its excellent performance under complex noise interference conditions.