改进SE-CAE去噪与两阶段域自适应的电机轴承故障诊断方法
By using an improved SE-CAE network and a two-stage domain adaptive training strategy, combined with multiple loss functions and domain-specific batch normalization, the problem of fault diagnosis of motor bearings under strong mixed noise and variable operating conditions is solved, achieving high-precision and robust fault identification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NORTHWESTERN POLYTECHNICAL UNIV
- Filing Date
- 2026-06-22
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies struggle to achieve high-precision and robust diagnosis of motor bearing faults under strong mixed noise and variable operating conditions. Signal processing methods are prone to losing weak fault features, deep learning models are poorly designed for noisy scenarios, and domain adaptive methods are prone to intra-class dispersion and negative transfer.
An improved compressed-excited convolutional autoencoder (SE-CAE) network is used for denoising and enhancement. Combined with a two-stage domain adaptive training strategy, a domain-specific batch normalized domain adaptive diagnostic network is constructed by training through a joint loss function of temporal reconstruction loss, sparse constraint loss and frequency domain consistency loss, so as to achieve high-precision identification of motor bearing faults.
It significantly improves the accuracy and robustness of motor bearing fault diagnosis under strong mixed noise and variable operating conditions, effectively preserves weak fault characteristics, and solves the diagnostic difficulties caused by noise interference and operating condition drift in the existing technology.
Smart Images
Figure CN122412784A_ABST