Study on Rapid Quantitative Analysis Method of Methanol Content in Methanol Gasoline by Raman Spectroscopy and Partial Least Squares

Mao-gang LI; Chun-hua YAN; Yao DU; Tian-long ZHANG; Hua LI

doi:10.3964/j.issn.1000-0593(2021)07-2099-06

Journals >Spectroscopy and Spectral Analysis >Volume 41 >Issue 7 >Page 2099 > Article

Spectroscopy and Spectral Analysis
Vol. 41, Issue 7, 2099 (2021)

Study on Rapid Quantitative Analysis Method of Methanol Content in Methanol Gasoline by Raman Spectroscopy and Partial Least Squares

Mao-gang LI^1、*, Chun-hua YAN^2、2;, Yao DU^1、1;, Tian-long ZHANG^2、2;, and Hua LI^{1、1; 2; *;}

Author Affiliations

¹1. College of Chemistry and Chemical Engineering, Xi'an Shiyou University, Xi'an 710065, China

²2. Key Laboratory of Synthetic and Natural Functional Molecular of the Ministry of Education, College of Chemistry & Materials Science, Northwest University, Xi'an 710127, China

show less

DOI: 10.3964/j.issn.1000-0593(2021)07-2099-06 Cite this Article

Mao-gang LI, Chun-hua YAN, Yao DU, Tian-long ZHANG, Hua LI. Study on Rapid Quantitative Analysis Method of Methanol Content in Methanol Gasoline by Raman Spectroscopy and Partial Least Squares[J]. Spectroscopy and Spectral Analysis, 2021, 41(7): 2099 Copy Citation Text

show less

Fig. 1. Raman spectra of different samples
(a): Methanol; (b): Gasoline;(c): Methanol gasoline with different methanol content

Download full size | View in the Article

Fig. 2. Prediction performance curve of PLS calibration model based on 5-flod CV with different LVs

Download full size | View in the Article

Fig. 3. Optimization figure of VIP variable importance threshold based on PLS calibration model
(a): Calibration set; (b): Prediction set

Download full size | View in the Article

Fig. 4. Prediction performance of three PLS calibration models based on Raman spectrum
(a): RAW-PLS; (b): WT-PLS; (c): VIP-PLS

Download full size | View in the Article

编号	体积分数/%	编号	体积分数/%	编号	体积分数/%	编号	体积分数/%	编号	体积分数/%
1	1.00	11	6.00	21	17.00	31	27.00	41	37.00
2	1.50	12^*	8.00	22^*	18.00	32^*	28.00	42^*	38.00
3^*	2.00	13	9.00	23	19.00	33	29.00	43	39.00
4	2.50	14	10.00	24	20.00	34	30.00	44	40.00
5	3.00	15^*	11.00	25^*	21.00	35^*	31.00	45^*	41.00
6^*	3.50	16	12.00	26	22.00	36	32.00	46	42.00
7	4.00	17	13.00	27	23.00	37	33.00	47	43.00
8	4.50	18^*	14.00	28^*	24.00	38^*	34.00	48^*	44.00
9^*	5.00	19	15.00	29	25.00	39	35.00	49	45.00
10	5.50	20	16.00	30	26.00	40	36.00

Table 1. Methanol volume fraction in methanol gasoline samples

View in the Article

预处理方法	5-flod CV		预测集
预处理方法	$R_{CV}^{2}$	RMSECV/%	$R_{P}^{2}$	RMSEP/%
RAW	0.967 4	0.034 5	0.936 2	0.041 3
MSC	0.967 4	0.034 4	0.955 1	0.035 0
SNV	0.969 1	0.033 6	0.953 4	0.033 7
D1st	0.965 7	0.035 4	0.906 8	0.050 7
Normalization	0.958 7	0.038 8	0.955 2	0.033 3
WT	0.974 3	0.030 7	0.960 4	0.034 1

Table 2. Comparison of prediction performance of PLS calibration models based on different spectral pretreatment methods

View in the Article

模型	建模时间/s	变量个数	5-flod CV		预测集
模型	建模时间/s	变量个数	$R_{CV}^{2}$	RMSECV /%	$R_{P}^{2}$	RMSEP /%
RAW-PLS	1.36	1 044	0.967 4	0.034 5	0.936 2	0.041 3
WT-PLS	1.32	1 044	0.974 3	0.030 7	0.960 4	0.034 1
VIP-PLS	0.28	22	0.973 7	0.031 1	0.960 4	0.034 1

Table 3. Comparison of prediction performance of the PLS calibration models based on different input variables

Download Citation

Save the article for my favorites

Paper Information