Rapid Detection of Total Organic Carbon in Oil Shale Based on Near Infrared Spectroscopy

Quan-lun LI; Zheng-guang CHEN; Xian-da SUN

doi:10.3964/j.issn.1000-0593(2022)06-1691-07

Journals >Spectroscopy and Spectral Analysis >Volume 42 >Issue 6 >Page 1691 > Article

Spectroscopy and Spectral Analysis
Vol. 42, Issue 6, 1691 (2022)

Rapid Detection of Total Organic Carbon in Oil Shale Based on Near Infrared Spectroscopy

Quan-lun LI^1、*, Zheng-guang CHEN^{1、1; *;}, and Xian-da SUN^2、2;

Author Affiliations

¹1. College of Information and Electrical Engineering, Heilongjiang Bayi Agricultural University, Daqing 163319, China

²2. Key Laboratory of Continental Shale Hydrocarbon Accumulation and Efficient Development, Ministry of Education, Northeast Petroleum University, Daqing 163318, China

show less

DOI: 10.3964/j.issn.1000-0593(2022)06-1691-07 Cite this Article

Quan-lun LI, Zheng-guang CHEN, Xian-da SUN. Rapid Detection of Total Organic Carbon in Oil Shale Based on Near Infrared Spectroscopy[J]. Spectroscopy and Spectral Analysis, 2022, 42(6): 1691 Copy Citation Text

show less

Fig. 1. Original spectra

Download full size | View in the Article

Fig. 2. Spectra after DT+BSC pretreatment

Download full size | View in the Article

Fig. 3. RMSE values of models with different characteristic wavelength numbers

Download full size | View in the Article

Fig. 4. Characteristic wavelengths selected based on SPA algorithm

Download full size | View in the Article

Fig. 5. Stability index of each wavelength variable and random variable

Download full size | View in the Article

Fig. 6. Characteristic wavelengths selected based on UVE algorithm

Download full size | View in the Article

Fig. 7. Process of selecting characteristic wavelengths based on CARS algorithm

Download full size | View in the Article

Fig. 8. Characteristic wavelengths selected based on CARS algorithm

Download full size | View in the Article

剔除方法	样本个数	$R_{c}^{2}$	RMSECV	$R_{v}^{2}$	RMSEP
未剔除	230	0.660 8	0.419 0	0.624 6	0.443 6
MCCV	216	0.718 9	0.366 3	0.688 4	0.386 3

Table 1. Modeling results before and after removing outlier samples

View in the Article

光谱预处理方法	$R_{c}^{2}$	RMSEC	$R_{v}^{2}$	RMSEV
无	0.718 9	0.366 3	0.688 4	0.386 3
DT	0.746 0	0.341 4	0.621 6	0.382 0
Derivative-1	0.706 7	0.303 8	0.643 6	0.310 6
Derivative-2	0.719 8	0.365 8	0.663 9	0.342 3
SNV	0.548 7	0.464 2	0.376 0	0.426 5
S-G	0.729 3	0.359 6	0.695 3	0.382 4
BSC	0.658 5	0.403 8	0.570 8	0.449 4
DT+SNV	0.739 3	1.352 7	0.578 0	0.334 8
DT+BSC	0.759 1	0.315 9	0.734 8	0.357 3

Table 2. Modeling results under different pretreatment methods

View in the Article

样本类型	样本数	最小值/%	最大值/%	均值/%	标准差
总体样本	216	0.053	3.922	1.835	0.693
校正集	144	0.053	3.922	1.897	0.747
验证集	72	0.699	2.783	1.713	0.551

Table 3. Sample set division

View in the Article

建模方法	特征波段筛选方法	波段数	校正集		验证集
建模方法	特征波段筛选方法	波段数	$R_{c}^{2}$	RMSEC	$R_{v}^{2}$	RMSEV
SVM	全谱波段	1 265	0.867 8	0.260 5	0.793 0	0.286 8
	CARS	65	0.912 4	0.204 6	0.906 6	0.222 0
	UVE	253	0.882 8	0.222 6	0.806 5	0.282 0
	SPA	16	0.883 5	0.244 0	0.815 3	0.278 0
RF	全谱波段	1 265	0.785 9	0.274 6	0.717 4	0.342 7
	CARS	65	0.812 5	0.297 6	0.765 9	0.316 9
	UVE	253	0.857 2	0.285 1	0.799 3	0.278 0
	SPA	16	0.842 5	0.293 7	0.812 7	0.323 5
PLS	全谱波段	1 265	0.765 5	0.314 8	0.742 1	0.342 5
	CARS	65	0.803 8	0.322 4	0.761 6	0.280 9
	UVE	253	0.834 5	0.284 3	0.754 1	0.310 0
	SPA	16	0.774 8	0.316 2	0.762 7	0.320 6

Table 4. Modeling results of PLS, RF and SVM

Quan-lun LI, Zheng-guang CHEN, Xian-da SUN. Rapid Detection of Total Organic Carbon in Oil Shale Based on Near Infrared Spectroscopy[J]. Spectroscopy and Spectral Analysis, 2022, 42(6): 1691

Download Citation

Tools

Save the article for my favorites

Paper Information