Chinese Journal of Quantum Electronics, Volume. 41, Issue 3, 553(2024)
Lithology analysis of rock with high repetition frequency laser⁃induced breakdown spectroscopy combined with convolutional neural network
Figures & Tables(21)
Fig. 4. Spectral line calibration of rock sample (No.1/D03) (The intensity of the calibrated lines≥2000)
Table 1. Information of rock samples
View tableView in ArticleTable 1. Information of rock samples
Number Name Origin Lithology Category A Category B 1 D03 Great Khingan coarse sandstone A5 B1 2 D04 Great Khingan medium sandstone A5 B2 3 D06 Great Khingan medium-coarse sandstone A5 B3 4 D08 Great Khingan coarse sandstone A5 B1 5 D10 Great Khingan medium-coarse sandstone A5 B3 6 P1-5 Great Khingan coarse sandstone A5 B1 7 P1-7 Great Khingan sandstone A5 B4 8 P1-11 Great Khingan conglomerate-medium-
coarse sandstone
A4 B5 9 P1-13-1 Great Khingan andesite A3 B6 10 P1-13-2 Great Khingan andesite A3 B6 11 P1-13-3 Great Khingan andesite A3 B6 12 P1-13-4 Great Khingan andesite A3 B6 13 P1-13-5 Great Khingan andesite A3 B6 14 19DJ-1 Shuangyang City conglomerate sandstone A1 B7 15 19DJ-4 Shuangyang City sandstone A5 B8 16 19DJ-5 Shuangyang City miscellaneous sandstone A2 B9
Table 2. Confusion matrix of dichotomous problem
View tableView in ArticleTable 2. Confusion matrix of dichotomous problem
True value Predicted value 1 0 1 TP FN 0 FP TN
Table 3. 1D⁃CNN result for rock 5 classification
View tableView in ArticleTable 3. 1D⁃CNN result for rock 5 classification
Label Total number Correct number Accuracy/% Overall accuracy/% A1 84 82 97.6 98.86 A2 63 62 98.4 A3 73 73 100 A4 70 70 100 A5 60 59 98.3
Table 4. 1D⁃CNN result for rock 9 classification
View tableView in ArticleTable 4. 1D⁃CNN result for rock 9 classification
Label Total number Correct number Accuracy/% Overall accuracy/% B1 54 27 50 93.02 B2 69 69 100 B3 75 64 85.3 B4 69 69 100 B5 76 73 96.1 B6 64 64 100 B7 76 76 100 B8 78 77 98.7 B9 69 67 97.1
Table 5. ResNet34 result for rock 5 classification
View tableView in ArticleTable 5. ResNet34 result for rock 5 classification
Label Total number Correct number Accuracy/% Overall accuracy/% A1 84 84 100 99.43 A2 63 62 98.4 A3 73 73 100 A4 70 70 100 A5 60 59 98.3
Table 6. ResNet34 result for rock 9 classification
View tableView in ArticleTable 6. ResNet34 result for rock 9 classification
Label Total number Correct number Accuracy/% Overall accuracy/% B1 54 46 85.2 97.14 B2 69 69 100 B3 75 69 92 B4 69 69 100 B5 76 74 97.4 B6 64 64 100 B7 76 75 98.7 B8 78 78 100 B9 69 68 98.6
Table 7. Comparative analysis of experimental results
View tableView in ArticleTable 7. Comparative analysis of experimental results
Categorization 1D-CNN ResNet34 Classification 5 98.86% 99.43% Classification 9 93.02% 97.14%
Tools
Get Citation
Copy Citation Text
Miao YANG, Ye ZHAN, Yuting FU, Guang YANG. Lithology analysis of rock with high repetition frequency laser⁃induced breakdown spectroscopy combined with convolutional neural network[J]. Chinese Journal of Quantum Electronics, 2024, 41(3): 553
Paper Information
Category: Special Issue on Key Technologies and Applications of LIBS
Received: Nov. 30, 2023
Accepted: --
Published Online: Jul. 17, 2024
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20