Weihan Li

Weihan Li


Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 November 2023 in Electronics
Verified
Journal Article
Open access
PDF file
GYGao YongShun
Gao YongShun
JLJie Liu
Jie Liu
WLWeihan Li
Weihan Li
MHMing Hou
Ming Hou
YLYang Li
Yang Li
HZHuimin Zhao
Huimin Zhao
In recent years, deep neural networks have shown superior performance in various fields, but interpretability has always been the Achilles’ heel of deep neural networks. The existing visual interpretation methods for deep neural networks still suffer from inaccurate and insufficient target localization and low-resolution saliency maps. To address the above issues, this paper presents a saliency map generation method based on image geometry augmentation and super-resolution called augmented high-order gradient weighting class activation mapping (augmented grad-CAM++). Unlike previous approaches that rely on a single input image to generate saliency maps, this method first introduces the image geometry augmentation technique to create a set of augmented images for the input image and generate activation mappings separately. Secondly, the augmented activation mappings are combined to form the final saliency map. Finally, a super-resolution technique is introduced to add pixel points to reconstruct the saliency map pixels to improve the resolution of the saliency map. The proposed method is applied to analyze standard image data and industrial surface defect images. The results indicate that, in experiments conducted on standard image data, the proposed method achieved a 3.1% improvement in the accuracy of capturing target objects compared to traditional methods. Furthermore, the resolution of saliency maps was three times higher than that of traditional methods. In the application of industrial surface defect detection, the proposed method demonstrated an 11.6% enhancement in the accuracy of capturing target objects, concurrently reducing the false positive rate. The presented approach enables more accurate and comprehensive capture of target objects with higher resolution, thereby enhancing the visual interpretability of deep neural networks. This improvement contributes to the greater interpretability of deep learning models in industrial applications, offering substantial performance gains for the practical deployment of deep learning networks in the industrial domain.

29 September 2021 in Applied Sciences
Verified
Journal Article
Open access
PDF file
WLWeihan Li
Weihan Li
YLYang Li
Yang Li
LYLing Yu
Ling Yu
JMJian Ma
Jian Ma
LZLei Zhu
Lei Zhu
Wu DengWu Deng
Wu Deng
A rolling element signal has a long transmission path in the acquisition process. The fault feature of the rolling element signal is more difficult to be extracted. Therefore, a novel weak fault feature extraction method using optimized variational mode decomposition with kurtosis mean (KMVMD) and maximum correlated kurtosis deconvolution based on power spectrum entropy and grid search (PGMCKD), namely KMVMD-PGMCKD, is proposed. In the proposed KMVMD-PGMCKD method, a VMD with kurtosis mean (KMVMD) is proposed. Then an adaptive parameter selection method based on power spectrum entropy and grid search for MCKD, namely PGMCKD, is proposed to determine the deconvolution period T and filter order L. The complementary advantages of the KMVMD and PGMCKD are integrated to construct a novel weak fault feature extraction model (KMVMD-PGMCKD). Finally, the power spectrum is employed to deal with the obtained signal by KMVMD-PGMCKD to effectively implement feature extraction. Bearing rolling element signals of Case Western Reserve University and actual rolling element data are selected to prove the validity of the KMVMD-PGMCKD. The experiment results show that the KMVMD-PGMCKD can effectively extract the fault features of bearing rolling elements and accurately diagnose weak faults under variable working conditions.