In Situ Detection Of Chemical Oxygen Demand In Reactive Dyeing Wastewater Based On UV/Vis Spectroscopy

The wastewater discharged from printing and dyeing processing consumes a large proportion of total water,results in environmental pollution.In this paper,an in-situ detection method of printing and dyeing wastewater COD and the influencing factors were systematically investigated based on Beer-Lambert law,chemometrics method and the self-developed optical fiber sensor system.The causes of the printing and dyeing wastewater COD,the relations between COD value and UV-Vis spectroscopy were systematically explored.The results provided valuable information for both theory study and practice application,and showed good potential to be applied in on-line water quality monitoring.In this study,by the optical fiber sensing system.the UV-Vis spectra of mixed dye solution samples were on-line tested.And COD SIMCA pattern recognition model was established to qualitatively measure COD,The results showed that the SIMCA model had better prediction effect for the samples.The correct rate was above 95%,and the recognition rate above 75%.By adjusting the optical path of the equipment,without dilution,high COD concentration of mixed dye solution of UV-Vis spectra can be directly detected.Using standardized,smooth and the first derivative of the spectral data preprocessing,principal component regression(PCR)and partial least squares regression(PLS)COD prediction model were established.The results showed that for the dye solution samples,there were high correlations and small noise between COD and spectra.Using the standard spectral pretreatment would be better than that of the other pretreatment methods.Both PCR and PLS models had high prediction precision,and the best results was PLS model(R2=0.874,RMSEP=19.6).Through the study of the actual printing and dyeing wastewater COD on-line quantitative detection,COD prediction model was established based on the boosting PLS,and the uninformative variable elimination algorithm(UVE)step was added in the boosting PLS algorithm.The prediction results(R2 = 0.8715 RMSEP=34.3)showed that the model had good prediction performance.Compared with PCR and PLS models,the computing time of boosting PLS increased,and the prediction accuracy improved greatly.Using UVE algorithm to reduce the number of input variables,the variables in the spectral variables decreased to 41.1%.Both the complexity of model and the modeling time were reduced.The results showed that the boosting PLS modeling using the UVE algorithm can realize online detection of printing and dyeing wastewater in situ rapid and accurate COD value.Through the analysis of turbidity effects on UV visible spectra for these samples:turbidity standard solution,the mixture of printing and dyeing wastewater after filtration and turbidity standard solution,and the actual printing and dyeing wastewater,turbidity interference detection of dyeing wastewater COD was studied.Under the effects of turbidity,for dyeing wastewater COD prediction accuracy of the model,different spectral pretreatment would result in different effects.The results showed that the turbidity standard solution and mixed solution had similar effects on the spectral signal,which made the ultraviolet spectrum especially 300nm the following areas had a greater offset.With the increase of the wavelength,especially in visible band,turbidity effect weakened,but there were some differences between different wavelengths of turbidity on the absorbance.Not by single wavelength method to eliminate the influence of turbidity and turbidity compensation.By using of multiple scattering corrections pretreatment,from the spectra and model error calculation,it was found to basically eliminate the influence of turbidity on the COD.The prediction results(R2 =0.843,RMSEP=36.7)showed that in the presence of turbidity interference,multiple scatter correction can improve the dyeing wastewater COD boosting PLS model prediction accuracy.According to the actual printing and dyeing wastewater as the research object,the boosting PLS prediction models in the ultraviolet,visible and full band were established.And the effects of temperature and pH on the printing and dyeing wastewater COD on-line detection were studied.The results showed that without the effects of other factors,the model in the UV wavelength(240-380nm)had minimum error and the highest correlation coefficient,R2 = 0.873,RMSEP = 34.3.According to the analysis results of the three bands models affected by temperature,when the temperature was below 40 ?,the printing and dyeing wastewater COD was little affected by temperature,and temperature effect can be neglected;At higher temperature,using visible band model,the temperature compensation function had higher correlation coefficient,R2 = 0.905.According to the analysis results of the three bands models affected by pH,when pH was less than 12,the UV band model was little affected by pH value.And the effects of pH can be ignored.Comprehensive analysis of effects of temperature and pH,the influence of the predicted COD calculated by the full band model by temperature was close to that of the ultraviolet band model,the maximum offset of COD affected by pH 7.77%was obtained.In the case of low precision,the pH effect can be ignored.