Search

Since minimum correlated generalized Lp/Lq deconvolution (MCG-Lp/Lq-D) cannot simultaneously have a good sensing ability to repetitive fault impulses and a strong robustness to noise when processing low SNR signals, a new deconvolution method, minimum improved correlated generalized Lp/Lq deconvolution (MICG-Lp/Lq-D), is proposed in this article. The proposed method is designed based on a novel norm, the improved correlated generalized Lp/Lq (ICG-Lp/Lq) norm. The ICG-Lp/Lq norm not only has the low-order statistics form that is similar to the correlated generalized Lp/Lq (CG-Lp/Lq) norm used in MCG-Lp/Lq-D, but also has the high shift-order signal correlation construction similar to the correlated kurtosis (CK) norm used in maximum correlated kurtosis deconvolution (MCKD). This guarantees that the proposed norm is robust enough to strong noise and can improve its sensing ability for repetitive impulses simultaneously. This feature of the ICG-Lp/Lq norm ensures that MICG-Lp/Lq-D can suppress outliers and extract weak impacts when processing low SNR signals. Simulated and experimental case studies verify the performance of the proposed method. The results indicate that the proposed method is robust enough to strong noise and very sensitive to repetitive impulses simultaneously, which indeed overcomes the aforementioned defects of MCG-Lp/Lq-D. Therefore, MICG-Lp/Lq-D can be considered a promising tool for detecting repetitive impulses. [ABSTRACT FROM AUTHOR]

Transfer learning (TL) has made great progress in intelligent fault diagnosis of bearing. However, due to the harsh working conditions of bearings in engineering practice, in addition to the bearing vibration signals, sensors also inevitably collect noise signals, which leads to the performance degradation of TL. In this paper, a novel model based on TL is proposed to solve the problem. Different from previous studies, an adaptive input length model instead of a fixed input length is established to diagnose bearings with different parameters. A signal processing method combining wide kernels in the first convolutional layer and pooling layer is used to feature denoise the input. Group convolutional and instance normalization based on 1D-CNN are constructed for feature extraction, health conditions classification, and denoise. An optimization objective function based on maximum mean discrepancy is introduced to align the feature distribution discrepancy. The model does not rely on any bearing data label information and realizes unsupervised TL. In the meanwhile, seven experiments including TL model and anti-noise model are conducted to compare and analyze the performance of intelligent fault diagnosis. Experimented on the axlebox bearing dataset of High-speed train, the accuracy is 85.98%, 93.62%, and 99.82% under SNR equal -6 dB, -4 dB, and -2 dB respectively. Experimented between bearing in electromechanical drive systems and high-speed train axlebox bearing dataset, the accuracy of the proposed method is 77.56%, 79.33%, and 96.22% under SNR equal -6 dB, -4 dB, and -2 dB respectively. In the meanwhile, the accuracy of the three advanced TL models and four state-of-the-art anti-noise models are below 50 %. The results indicate that the proposed model has good bearing fault diagnosis ability for TL different datasets under noisy conditions. • Adaptive input length strategy is applied to unsupervised TL bearing diagnosis. • The proposed model has good anti-noise ability even under -6 dB for cross-domain TL. • All the hyperparameters are learned by itself. • Seven comparison experiments including TL and anti-noise model are conducted. [ABSTRACT FROM AUTHOR]