Study on Identification Method Based on XGBoost Model for Aluminum Alloy Using Laser-Induced Breakdown Spectroscopy

Chen-yang LI; Xiong-fei CHEN; Yong ZHANG; Ya-wen WANG; Zhong-chao TIAN; Shi-gong WANG; Zhen-yang ZHAO; Ying LIU; Peng-yu LIU

doi:10.3964/j.issn.1000-0593(2021)02-0624-05

Spectroscopy and Spectral Analysis, Volume. 41, Issue 2, 624(2021)

Study on Identification Method Based on XGBoost Model for Aluminum Alloy Using Laser-Induced Breakdown Spectroscopy

Chen-yang LI^1,1、*, Xiong-fei CHEN^1,1, Yong ZHANG^1,1, Ya-wen WANG^1,1, Zhong-chao TIAN^1,1, Shi-gong WANG^1,1, Zhen-yang ZHAO^1,1, Ying LIU^1,1, and Peng-yu LIU^1,1

Author Affiliations

¹[in Chinese]

¹1. National Analysis and Testing Center of Nonferrous Metals and Electronic Materials, GRINM Group Corporation Limited, Beijing 100088, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(9)

Fig. 1. The schematic diagram of LIBS experiment setup

Download full size

View in Article

Fig. 2. The spectral matrices of six kinds of standard aluminium alloy

Download full size

View in Article

Fig. 3. The feature importance scores given by XGBoost

Download full size

View in Article

Fig. 4. Distribution map of spectral data under 4 featres of XGBoost model

Download full size

View in Article

Table 1. The values concentration (Wt.%) of impurity elements in six alloys

View table

View in Article

Table 1. The values concentration (Wt.%) of impurity elements in six alloys

编号No.	Si	Fe	Cu	Mn	Mg	Cr	Ni	Zn	Ti
1^#	0.541	0.206	0.040	0.864	1.170	0.035	0.059	0.297	0.114
2^#	1.010	0.363	0.228	0.522	0.693	0.161	0.043	0.226	0.055
3^#	1.240	0.428	0.487	0.233	0.933	0.354	0.027	0.094	0.026
4^#	0.235	0.697	0.705	1.180	0.411	0.218	0.089	0.166	0.066
5^#	1.580	0.848	0.078	1.480	0.068	0.288	0.115	0.041	0.008
6^#	0.052	0.074	0.927	0.042	1.460	0.039	0.013	0.361	0.160

Table 2. Spectral lines for analysis
View table
View in Article
Table 2. Spectral lines for analysis
分析元素分析谱线(NIST)/nm
Si 250.69, 251.61
Fe 259.94
Cu 324.75, 327.40
Mn 259.37
Cr 357.87
Ni 232.00
Zn 213.86
Ti 334.94

Table 3. Super parameters of XGBoost model
View table
View in Article
Table 3. Super parameters of XGBoost model
超参数取值
迭代模型 gbtree
损失函数 Multi∶Softmax
学习率 0.1
树的深度 6
L₂正则化参数 2
类别数 6

Table 4. The Coefficient of Variation of all analytical elements
View table
View in Article
Table 4. The Coefficient of Variation of all analytical elements
分析元素变异系数
Cu 0.871 2
Ti 0.792 2
Mn 0.772 1
Si 0.770 4
Cr 0.712 8
Fe 0.670 4
Ni 0.668 1
Mg 0.643 2

Table 5. Identification results of Test Sets
View table
View in Article
Table 5. Identification results of Test Sets
1^# 2^# 3^# 4^# 5^# 6^# 准确率/%
1^# 36 2 94.73
2^# 22 1 95.65
3^# 2 26 1 89.66
4^# 35 100
5^# 27 100
6^# 28 100
总体 96.67

Tools

Get Citation

Copy Citation Text

Chen-yang LI, Xiong-fei CHEN, Yong ZHANG, Ya-wen WANG, Zhong-chao TIAN, Shi-gong WANG, Zhen-yang ZHAO, Ying LIU, Peng-yu LIU. Study on Identification Method Based on XGBoost Model for Aluminum Alloy Using Laser-Induced Breakdown Spectroscopy[J]. Spectroscopy and Spectral Analysis, 2021, 41(2): 624

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites