Abstract
As a critical component of the drive motor in electric heavy trucks (EHTs), bearings provide power for the operation of the EHTs. However, current fault diagnosis methods for EHTs are scarce and face challenges such as continuous variable speed conditions and strong background noise interference during diagnosis. To address these issues, this study proposes a fault diagnosis method for EHTs drive motor bearings based on multidimensional chromatic symmetrized dot pattern (MC-SDP). First, rotational speed signals are extracted from the time–frequency spectrum of current signals. Second, order analysis is employed to decouple rotational speed variations from vibration signals, thereby eliminating their impact on fault diagnosis accuracy. Then, improve the SDP by using multidimensional vibration signals and taking the standard deviation as a color feature for feature extraction, addressing the issue of unclear fault characteristics under strong background noise interference, thus enhancing fault diagnosis efficiency. Finally, the squeeze-and-excitation (SE) module is added to the convolutional neural network (CNN) to enhance its sensitivity to color features, and based on the SE-CNN, fault pattern recognition is performed on the MC-SDP images obtained through feature extraction. Experimental results on vibration data from a drive motor testbed demonstrate a fault diagnosis accuracy of 97.5%, outperforming traditional methods in diagnostic performance.
Funding Information
  • Innovation Funds Plan of Henan University of Technology (No. 2021ZKCJ07)
  • Henan Provincial Technology Deputy Directors Program

This publication has 43 references indexed in Scilit: