Acta Photonica Sinica, Volume. 51, Issue 4, 0430001(2022)
Rapid Classification of Laser Induced Breakdown Spectroscopy of Titanium Alloys
Figures & Tables(14)
Fig. 3. The change of peak intensity and signal to noise ratio with unit pulse energy
Fig. 4. The change of peak intensity and signal to noise ratio with trigger delay
Fig. 6. Confusion matrix of training set of original spectrums after weighted KNN classification and confusion matrix of test set classification result
Fig. 7. Confusion matrix of weighted KNN classification for spectral training set after data processing and confusion matrix of test set classification result
Fig. 9. Training set confusion matrix of final classification result and confusion matrix of test set classification result
Table 1. Titanium alloy number and element content
View tableView in ArticleTable 1. Titanium alloy number and element content
Material Number Concentration of element/% Al V Fe Si C Ti TC4-1 GBW02503 3.90 5.56 0.39 0.28 0.16 89.62 TC4-2 GBW02504 4.67 5.01 0.31 0.20 0.12 89.69 TC4-3 GBW02505 5.38 3.41 0.24 0.12 0.10 90.75 TC4-4 GBW02507 6.78 3.85 0.13 0.09 0.02 89.13 TC4KY - 6.24 4.08 0.05 0.02 0.01 89.60
Table 2. Spectral wavelength，transition probability and energy level
View tableView in ArticleTable 2. Spectral wavelength，transition probability and energy level
Element Wavelength/nm Ak/S-1 E1/eV E2/eV Al I 394.401 4.99×107 0 3.14 Al I 396.152 9.85×107 0 3.14 V II 311.071 1.58×108 0.35 4.33 V II 355.680 6.40×107 1.13 4.61 Ti I 395.820 4.88×107 0.05 3.18 Ti I 399.864 4.81×107 0.05 3.15
Table 3. Characteristic lines for spectral screening
View tableView in ArticleTable 3. Characteristic lines for spectral screening
Element Al I Al II V I V II Fe II Wavelength/nm 226.90
396.15
281.62 411.15
439.55
268.87
310.23
319.07
238.24
259.99
Table 4. Comparison of results between different algorithms
View tableView in ArticleTable 4. Comparison of results between different algorithms
K-means Decision tree NBC SVM KNN Training time/s 43.86 79.28 78.95 91.18 83.91 Training set cross validation accuracy/% 53.34 81.6 97.9 98.4 98.64 Test set classification accuracy/% 58.31 82.91 96.79 98.5 99.14
Table 5. Comparison of optimization results
View tableView in ArticleTable 5. Comparison of optimization results
Raw data Data processing KNN model optimization Training time/s 1 232.41 84.79 83.91 Training set cross validation accuracy/% 81.40 94.00 98.64 Test set classification accuracy/% 84.20 95.92 99.14
Tools
Get Citation
Copy Citation Text
Cheng XU, Fang LI, Feng CHEN, Deng ZHANG, Fan DENG, Lianbo GUO. Rapid Classification of Laser Induced Breakdown Spectroscopy of Titanium Alloys[J]. Acta Photonica Sinica, 2022, 51(4): 0430001
Paper Information
Category:
Received: Nov. 9, 2021
Accepted: Jan. 25, 2022
Published Online: May. 18, 2022
The Author Email: LI Fang (lifang@wit.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20