Design of Water Quality COD Online Detection System Based on Ultraviolet-Visible Spectroscopy

Jiacheng Cai; Xiao Deng; Chongxuan Kan; Shuang Zhao

doi:10.3788/LOP202259.0130001

Journals >Laser & Optoelectronics Progress >Volume 59 >Issue 1 >Page 0130001 > Article

Laser & Optoelectronics Progress
Vol. 59, Issue 1, 0130001 (2022)

Design of Water Quality COD Online Detection System Based on Ultraviolet-Visible Spectroscopy

Jiacheng Cai¹, Xiao Deng^1,2,*, Chongxuan Kan¹, and Shuang Zhao¹

Author Affiliations

¹College of Physics and Optoelectronics, Taiyuan University of Technology, Taiyuan , Shanxi 030024, China

²Key Laboratory of Advanced Transducers and Intelligent Control System, Ministry of Education and Shanxi Province, Taiyuan University of Technology, Taiyuan , Shanxi 030024, China

show less

DOI: 10.3788/LOP202259.0130001 Cite this Article Set citation alerts

Jiacheng Cai, Xiao Deng, Chongxuan Kan, Shuang Zhao. Design of Water Quality COD Online Detection System Based on Ultraviolet-Visible Spectroscopy[J]. Laser & Optoelectronics Progress, 2022, 59(1): 0130001 Copy Citation Text

show less

Fig. 1. Schematic diagram of the overall system structure

Download full size

Fig. 2. Absorption spectrum of COD standard solution. (a) Transmitted light intensity curve; (b) absorbance curve

Download full size

Fig. 3. Spectral data prediction regression model. (a) 220~245 nm band; (b) 260~285 nm band

Download full size

Fig. 4. COD concentration-voltage data model. (a) Normalized spectral curve; (b) voltage regression model; (c) voltage solution model

Download full size

Fig. 5. Absorption spectrum of mixed solution. (a) Absorbance curve of mixed solution with COD concentration of 50 mg/L and different turbidity; (b) absorbance curve of standard turbidity solution; (c) absorbance of the mixed solution minus the absorbance of the turbidity solution

Download full size

Fig. 6. Compensation model. (a) Rrelationship between K_N and wavelength under different turbidity; (b) compensated absorbance curve

Download full size

Fig. 7. Absorption spectra of 50 mg/L COD solution at different temperature. (a) Transmitted light intensity curve; (b) absorbance curve; (c) fitting equation at 250 nm wavelength

Download full size

COD/ （mg·L^-1）	220~245 nm wavelength PLS algorithm model		260~285 nm wavelength PLS algorithm model		Voltage solution model
COD/ （mg·L^-1）	Predictive value /（mg·L^-1）	Relative error /（mg·L^-1）	Predictive value /（mg·L^-1）	Relative error /（mg·L^-1）	Predictive value /（mg·L^-1）	Relative error / %
5	4.9118	-0.0178	5.2664	0.0533	5.1557	0.0311
10	10.6633	0.0663	9.8818	-0.0118	10.8115	0.0811
15	15.6599	0.0440	15.6939	0.0463	13.7623	-0.0825
20	19.9684	-0.0016	19.5969	-0.0202	20.8115	0.0406
25	24.9227	-0.0031	24.5523	-0.0179	24.9918	-0.0003
30	29.8816	-0.0039	29.3947	-0.0202	30.0738	0.0025
35	34.5830	-0.0119	42.4895	0.2140	35.0738	0.0021
40	40.7377	0.0184	40.4746	0.0119	38.2705	-0.0432
45	45.0068	0.0002	45.1604	0.0036	44.2541	-0.0166
50	50.1648	0.0033	42.4895	-0.1502	51.8771	0.0375

Table 1. Relative error of the predicted value of COD concentration under different algorithm models

Turbidity（NTU）	$K_{N} (λ)$	R₂
10	y = -0.0003966x² + 0.16503x-15.77201	0.99363
20	y = -0.00045692x² + 0.19838x-20.44448	0.99385
30	y = -0.00050976x² + 0.23691x-26.59873	0.99542
40	y = -0.00056239x² + 0.26968x-31.41166	0.99611
50	y = -0.00065806x² + 0.32548x-39.25635	0.99817

Table 2. Fitting formula of K_N(λ) under different turbidities

Characteristic wavelength /nm	Evaluation index			Characteristic wavelength（nm）	Evaluation index
Characteristic wavelength /nm	r	R₂	MSE	Characteristic wavelength（nm）	r	R₂	MSE
250	0.98650	0.97315	0.00008	275	0.98295	0.96619	0.00012
255	0.98640	0.97299	0.00009	280	0.98195	0.96422	0.00012
260	0.98580	0.97172	0.00009	285	0.98036	0.96111	0.00013
265	0.98500	0.97026	0.00010	290	0.97771	0.95591	0.00014
270	0.98430	0.96874	0.00010	295	0.97461	0.94986	0.00016