Quantitative Detection of Multi‑Component Rubber Additives Based on Terahertz Spectral Data Fusion

Xianhua Yin; Huicong Chen; Huo Zhang

doi:10.3788/CJL230807

Chinese Journal of Lasers, Volume. 51, Issue 5, 0514001(2024)

Quantitative Detection of Multi‑Component Rubber Additives Based on Terahertz Spectral Data Fusion

Xianhua Yin^1,2, Huicong Chen^1,2, and Huo Zhang^1,2、*

¹School of Electronic Engineering and Automation, Guilin University of Electronic Science and Technology, Guilin 541004, Guangxi , China

²Guangxi Key Laboratory of Automatic Detection Technology and Instrument, Guilin 541004, Guangxi , China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(12)

Fig. 1. Physical photo of the Z-3 system

Download full size

Fig. 2. Schematic diagram of THz system

Download full size

Fig. 3. Absorption spectra of five kinds of pure substances

Download full size

Fig. 4. Average absorption spectra for five-component mixtures

Download full size

Fig. 5. Absorption spectra of five-component mixtures

Download full size

$Predicted mass fraction of antioxidant MB with different models. (a) PLSR model; (b) SVR model$

Fig. 6. Predicted mass fraction of antioxidant MB with different models. (a) PLSR model; (b) SVR model

Download full size

Fig. 7. Predicted results of antioxidant MB. (a) Predicted results based on absorption spectra; (b) predicted results based on derivative spectra

Download full size

Fig. 8. Predicted results of antioxidant MB. (a) Low-level data fusion; (b) mid-level data fusion (SPA); (c) mid-level data fusion (MCUVE); (d) high-level data fusion

Download full size

Table 1. Five-component mixture content ratio
View table
Table 1. Five-component mixture content ratio
Number Mass fraction /%
NBR Silica ZnO Antioxidant H Antioxidant MB
1 60 10 6 0 24
2 60 10 6 3 21
3 60 10 6 6 18
4 60 10 6 9 15
5 60 10 6 12 12
6 60 10 6 15 9
7 60 10 6 18 6
8 60 10 6 21 3
9 60 10 6 24 0

Table 2. PLSR and SVR modeling results of five-component mixtures
View table
Table 2. PLSR and SVR modeling results of five-component mixtures
Model Training set Prediction set
R_c RMSEC /% R_p RMSEP /%
PLSR 0.9269 2.9108 0.9260 2.9947
SVR 0.9760 1.6899 0.9653 2.0219

Table 3. SVR modeling results of five-component mixtures
View table
Table 3. SVR modeling results of five-component mixtures
Spectrum Training set Prediction set
R_c RMSEC /% R_p RMSEP /%
Absorption 0.9760 1.6899 0.9653 2.0219
Derivative 0.9778 1.6249 0.9528 2.3523

Table 4. Data fusion modeling results of five-component mixtures

View table

Table 4. Data fusion modeling results of five-component mixtures

Model	Training set		Testing set
Model	R_c	RMSEC /%	R_p	RMSEP /%
Low-level data fusion	0.9908	1.0476	0.9701	1.8791
Mid-level data fusion（SPA）	0.9841	1.3759	0.9693	1.9040
Mid-level data fusion（MCUVE）	0.9900	1.0947	0.9758	1.6941
High-level data fusion	0.9872	1.2367	0.9732	1.7798

Tools

Get Citation

Copy Citation Text

Xianhua Yin, Huicong Chen, Huo Zhang. Quantitative Detection of Multi‑Component Rubber Additives Based on Terahertz Spectral Data Fusion[J]. Chinese Journal of Lasers, 2024, 51(5): 0514001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites