Acta Photonica Sinica, Volume. 53, Issue 6, 0630001(2024)
Adaptive Baseline Correction Method for Laser-induced Breakdown Spectroscopy
Figures & Tables(10)
Fig. 1. Generalized logic function of arPLS and Gaussian 3
Fig. 4. 6061 aluminum-based alloy LIBS corrected by different baseline correction methods
Fig. 5. The baselines fitted by the LIBS of 6061 aluminum-based alloy with different arPLS input parameters and the PSO-arPLS adaptive method， and the PSO-arPLS corrected spectrum
Fig. 6. Confusion matrix results of cubic kernel function SVM classification model after PSO-arPLS baseline correction
Table 1. Chemical composition of 6061， 6063 and 7075 aluminum-based alloys
View tableView in ArticleTable 1. Chemical composition of 6061， 6063 and 7075 aluminum-based alloys
Material Concentration of element/% Al Mg Si Cu Mn Zn Ti Fe Cr 6061 Remain 0.8~1.2 0.40~0.8 0.15~0.40 0.15 0.25 0.15 0.70 - 6063 Remain 0.45~0.9 0.20~0.6 0.10 0.10 0.10 0.10 0.35 0.10 7075 Remain 2.1~2.9 0.40 1.2~2.0 0.30 5.1~6.1 0.20 0.50 0.18~0.28
Table 2. SNR at Al I and Al II， Mg， Si， Ca， Na， and Zn by different baseline correction pretreatment methods
View tableView in ArticleTable 2. SNR at Al I and Al II， Mg， Si， Ca， Na， and Zn by different baseline correction pretreatment methods
Spectral processing SNR Mg 279.541 nm Ti 309.239 nm Al Ⅰ
394.145 nm
Al Ⅱ
395.953 nm
Ca
422.457 nm
Zn
472.074 nm
Na
589.745 nm
airPLS 76 83 121 135 123 91 323 PSO-arPLS 103 116 148 159 140 123 356 asPLS 82 100 123 140 125 112 322 Original spectrum 98 92 143 138 126 103 350
Table 3. Accuracy of SVM classification model for three LIBS data sets of aluminum base alloy standard samples processed by different baseline correction pretreatment methods
View tableView in ArticleTable 3. Accuracy of SVM classification model for three LIBS data sets of aluminum base alloy standard samples processed by different baseline correction pretreatment methods
Methods Cross-validation set/% Test set/% Original LIBS dataset 90.2 100 airPLS 92.3 88.6 asPLS 94.2 96.4 PSO-arPLS 99.8 100
Table 4. The TPR and FNR of the confusion matrix obtained by training SVM model with different kernel functions and kernel scaling modes with LIBS data of three Al-base alloy standard samples
View tableView in ArticleTable 4. The TPR and FNR of the confusion matrix obtained by training SVM model with different kernel functions and kernel scaling modes with LIBS data of three Al-base alloy standard samples
Kernel function Kernel scale mode TPR FNR Quadratic Auto 96.8% 3.2% Cubic Auto 99.8% 0.2% Gaussian Rough 94.9% 5.1% Medium 97.7% 2.3% Fine 84.1% 15.9%
Tools
Get Citation
Copy Citation Text
Hongyu YAN, Yu ZHAO, Yuanyuan CHEN, Hao LIU, Jie WANG, Rui ZHANG, Zhibin WANG. Adaptive Baseline Correction Method for Laser-induced Breakdown Spectroscopy[J]. Acta Photonica Sinica, 2024, 53(6): 0630001
Paper Information
Category:
Received: Nov. 21, 2023
Accepted: Dec. 18, 2023
Published Online: Jul. 16, 2024
The Author Email: Yu ZHAO (ciomzhaoyu@sina.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20