Infrared and Laser Engineering, Volume. 52, Issue 4, 20220492(2023)
Research on circuit board fault diagnosis based on infrared temperature series
Figures & Tables(14)
Fig. 7. Results of training and validation in Datasets_1 and Datasets_2
Fig. 8. Comparison of diagnostic results of different algorithm models
Table 1. Failure mode description
View tableView in ArticleTable 1. Failure mode description
Mode Description Mode Description Mode Description Mode Description F1 Normal F8 pin2 of U2-open F15 pin3 of U3-open F22 pin2 and 3 of U2 chip-short F2 pin2 of U1-open F9 pin3 of U2-open F16 pin4 of U3-open F23 pin3 and 4 of U2 chip-short F3 pin3 of U1-open F10 pin4 of U2-open F17 pin6 of U3-open F24 pin6 and 7 of U2 chip-short F4 pin4 of U1-open F11 pin5 of U2-open F18 pin7 of U3-open F25 pin2 and 3 of U3 chip-short F5 pin6 of U1-open F12 pin6 of U2-open F19 pin2 and 3 of U1 chip-short F26 pin3 and 4 of U3 chip-short F6 pin7 of U1-open F13 pin7 of U2-open F20 pin3 and 4 of U1 chip-short F27 pin6 and 7 of U3 chip-short F7 pin1 of U1-open F14 pin1 of U3-open F21 pin6 and 7 of U1 chip-short
Table 2. Parameters of self-made temperature series datasets
View tableView in ArticleTable 2. Parameters of self-made temperature series datasets
Samples Sequence Features Purpose Datasets_1 19548 120 3 Train & Verify Datasets_2 19548 120 6 Train & Verify Datasets_3 4887 120 3 Test Datasets_4 4887 120 6 Test
Table 3. Hyperparameters settings of FEN
View tableView in ArticleTable 3. Hyperparameters settings of FEN
Parameters DCB：Layer_1 DCB：Layer_2 DCB：Layer_3 CB Dilated_1 Dilated_2 Dilated_3 Dilated_1 Dilated_2 Dilated_3 Conv1D Block1 Block2 Block3 Filters 128 256 128 128 Dilation rate 1 2 4 1 2 4 1 - Activation LeakRelu Filter size 8 5 3 7 5 1 Receptive field 1×8 2×8 4×8 1×5 2×5 4×5 1×3 7 5 1 Scales proportion 1 1 2 1 1 2 - - 1 2 1 1 2 1 2 1 1 2 1 1
Table 4. Performance of different kernel quantity
View tableView in ArticleTable 4. Performance of different kernel quantity
1∶1∶2 1∶2∶1 2∶1∶1 Datasets_1 92.39% 95.15% 92.18% Datasets_2 98.19% 98.98% 97.07%
Table 5. Performance comparison of different classi-fication algorithm models
View tableView in ArticleTable 5. Performance comparison of different classi-fication algorithm models
FCN MFCN LSTM LSTM-FCN Proposed Datasets_1 93.21% 94.71% 91.78% 94.83% 95.15% Datasets_2 97.12% 96.69% 96.35% 98.44% 98.98% Datasets_3 80.71% 83.30% 80.09% 85.80% 91.15% Datasets_4 88.8% 89.56% 85.58% 91.71% 96.27%
Table 6. Performance comparison of ablation experiment
View tableView in ArticleTable 6. Performance comparison of ablation experiment
Relu No_Dilate No_SelfAtt AttLSTM 1_LSTM 3_LSTM Datasets_3 89.73% 90.23% 90.21% 89.37% 88.54% 83.38% Datasets_4 95.75% 95.67% 95.07% 95.97% 93.82% 90.80%
Tools
Get Citation
Copy Citation Text
Jianxin Hao, Li Wang. Research on circuit board fault diagnosis based on infrared temperature series[J]. Infrared and Laser Engineering, 2023, 52(4): 20220492
Paper Information
Category: Infrared technology and application
Received: Jul. 14, 2022
Accepted: --
Published Online: Jul. 4, 2023
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20