An Infrared Thermal Imaging Method for Circuit Board Component Troubleshooting
Xudong Song, Chao Liu, Liang Song, Jun Liu
(Dalian Jiaotong University, School of Intelligent Engineering, Dalian, Liaoning 116028)
Abstract: To detect internal faults in circuit board components, an infrared thermal imaging method for troubleshooting circuit board components is proposed. First, a network model called LGC-Net is introduced for fault diagnosis. To enhance the network’s ability to extract global features, a dual-layer attention module and a channel feature extraction module are incorporated into the base network ResNet50. Secondly, to further improve the accuracy of fault diagnosis, an optimized residual structure and a pre-trained model with fine-tuning are employed for constructing the diagnostic model. Finally, image enhancement is performed on the collected infrared thermal images, and experiments on the fault diagnosis model are conducted. The experiments show that the LGC-Net network model achieves a fault diagnosis accuracy of 98.92%, which is an improvement of 1.9 percentage points over the optimal accuracy of classical network fault diagnosis, while the diagnosis time for a single infrared thermal image is 246ms, reducing the shortest diagnosis time by 299ms compared to other models. This indicates that the proposed method can effectively identify faults in circuit board components through infrared thermal imaging, thereby improving the accuracy of fault diagnosis for circuit board components.
Keywords: Infrared thermal imaging; components; residual structure; image enhancement; fault diagnosis
DOI:10.13291/j.cnki.djdxac.2025.04.019
Citation format: Xudong Song, Chao Liu, Liang Song, et al. An Infrared Thermal Imaging Method for Circuit Board Component Troubleshooting[J]. Journal of Dalian Jiaotong University, 2025, 46(4): 155-160.
English citation format: XUDONG SONG, CHAO LIU, LIANG SONG, et al. An Infrared Thermal Imaging Method for Circuit Board Component Troubleshooting[J]. Journal of Dalian Jiaotong University, 2025, 46(4): 155-160.
Scan to view the full text
Journal official website
https://dltd.cbpt.cnki.net
Editor: Aijuan Yan
Typesetting: Sibing Liu
Review: Ruijun Guo