Detection of Chemical Oxygen Demand in Water Based on Multi-Spectral Fusion of Ultraviolet and Fluorescence

Kunpeng Zhou; Xufang Bai; Weihong Bi

doi:10.3788/LOP55.113003

Journals >Laser & Optoelectronics Progress >Volume 55 >Issue 11 >Page 113003 > Article

Laser & Optoelectronics Progress
Vol. 55, Issue 11, 113003 (2018)

Detection of Chemical Oxygen Demand in Water Based on Multi-Spectral Fusion of Ultraviolet and Fluorescence

Kunpeng Zhou¹, Xufang Bai¹, and Weihong Bi^2、*

Author Affiliations

¹ College of Physics and Electronic Information, Inner Mongolia University for Nationalities, Tongliao, Inner Mongolia 0 28000, China

² Key Laboratory for Special Fiber and Fiber Sensor of Hebei Province, School of Information Science and Engineering, Yanshan University, Qinhuangdao, Hebei 0 66004, China

show less

DOI: 10.3788/LOP55.113003 Cite this Article Set citation alerts

Kunpeng Zhou, Xufang Bai, Weihong Bi. Detection of Chemical Oxygen Demand in Water Based on Multi-Spectral Fusion of Ultraviolet and Fluorescence[J]. Laser & Optoelectronics Progress, 2018, 55(11): 113003 Copy Citation Text

show less

Fig. 1. UV absorption spectra of experimental water samples

Download full size

Fig. 2. Contour maps before and after removal of scattering of sea water samples. (a) Before removal of scattering; (b) after removal of scattering

Download full size

Fig. 3. Modeling process of the optimal model in this study

Download full size

Fig. 4. Flow chart of low level data fusion modeling based on UV and fluorescence multi-spectral information

Download full size

Fig. 5. Flow chart of mid-level data fusion modeling based on UV and fluorescence multi-spectral information

Download full size

Fig. 6. Prediction results of MLDF5-PSO-LSSVM model. (a) Calibration set; (b) validation set

Download full size

Featureextractionmethod	Optimalpreprocessingmethod	Preferredband /nm	Calibration set		Validation set
Featureextractionmethod	Optimalpreprocessingmethod	Preferredband /nm	$\begin{matrix} R_{c}^{2} \end{matrix}$	RMSECV /(mg·L^-1)	$\begin{matrix} R_{p}^{2} \end{matrix}$	RMSEP /(mg·L^-1)
ACO-iPLS	S-G smoothing &eliminationconstant offset	220.553,222.934,231.857,247.308,254.432,256.213,260.959,266.889,271.631,287.619	0.9983	0.6582	0.9839	2.0794

Table 1. Modeling results of UV spectra of experimental water samples by PSO-LSSVM

Excitationwavelength /nm	Calibration set		Validation set
Excitationwavelength /nm	$\begin{matrix} R_{c}^{2} \end{matrix}$	RMSECV /(mg·L^-1)	$\begin{matrix} R_{p}^{2} \end{matrix}$	RMSEP /(mg·L^-1)
220	0.9999	0.1004	0.9692	2.0987
225	0.9999	0.1703	0.9783	1.7783
230	0.9999	0.0468	0.9779	1.8459
235	0.9980	0.6351	0.9876	1.3725
240	0.9293	3.7334	0.9661	2.5271
245	0.9995	0.3051	0.9684	2.1227
250	0.9980	0.6545	0.9750	1.9570
255	0.9283	3.7744	0.9620	2.5084
260	0.9939	1.1336	0.9634	2.6259
265	0.9998	0.1789	0.9901	1.1986
270	0.9916	1.3183	0.9724	2.1348
275	0.9342	3.3273	0.9038	4.3444
280	0.9923	1.2832	0.9563	2.8420
285	0.9999	0.0486	0.9723	2.2662
290	0.9999	0.0480	0.9839	1.4133
295	0.9996	0.2965	0.9737	2.1070
300	0.9896	1.4185	0.9725	2.1539
305	0.9999	0.0912	0.9656	2.3591
310	0.9993	0.3665	0.9789	1.7811
315	0.9936	1.1119	0.9800	2.2497
320	0.9966	0.8122	0.9690	2.2926
325	0.9999	0.1123	0.9514	3.1902
330	0.9846	1.5396	0.9647	2.6610

Table 2. Modeling results of fluorescence spectra of experimental water samples by PSO-LSSVM

Fusion approach	Calibration set		Validation set
Fusion approach	$\begin{matrix} R_{c}^{2} \end{matrix}$	RMSECV /(mg·L^-1)	$\begin{matrix} R_{p}^{2} \end{matrix}$	RMSEP /(mg·L^-1)
LLDF1	0.9999	0.1361	0.9821	1.6552
LLDF2	0.9999	0.0585	0.9886	1.2582
LLDF3	0.9999	0.3969	0.9876	1.4495
LLDF4	0.9999	0.0450	0.9857	1.5243
LLDF5	0.9999	0.0377	0.9766	1.8591
LLDF6	0.9999	0.0713	0.9860	1.5294
LLDF7	0.9999	0.0352	0.9875	1.4589

Table 3. PSO-LSSVM modeling results of low level data fusion based on UV and fluorescence multi-spectral information

Fusion approach	Calibration set		Validation set
Fusion approach	$\begin{matrix} R_{c}^{2} \end{matrix}$	RMSECV /(mg·L^-1)	$\begin{matrix} R_{p}^{2} \end{matrix}$	RMSEP /(mg·L^-1)
MLDF1	0.9999	0.0284	0.9863	1.4265
MLDF2	0.9999	0.0285	0.9864	1.4265
MLDF3	0.9999	0.0409	0.9886	1.4086
MLDF4	0.9999	0.0168	0.9856	1.5116
MLDF5	0.9999	0.0304	0.9912	1.1297
MLDF6	0.9999	0.0164	0.9833	1.5715
MLDF7	0.9999	0.0131	0.9848	1.5517

Table 4. PSO-LSSVM modeling results of mid-level data fusion based on UV and fluorescence multi-spectral information

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information