Near Infrared Spectroscopy Analysis Of Nitrogen Removal From River Overlying Water By Artificial Aeration

The eutrophication and black odor of water are the serious consequences of the destruction of water environment in China,and the excess of nitrogen is one of the important causes of water pollution.At present,continuous aeration is usually used to solve this problem.However,continuous aeration not only has high energy consumption,but also can not provide anaerobic state,which couldn’t makes denitrification go smoothly.The thesis topic mainly focus on the treatment of the overlying water through artificial aeration method under the simulation of the river in the laboratory environment,which effectively improves the removal rate of nitrogen in water.And the rapid and non-destructive monitoring of total nitrogen,ammonia nitrogen,nitrate nitrogen and nitrite nitrogen concentration during the removal process by near infrared spectroscopy combined with stoichiometric method provides a theoretical basis for the practical application of this technology in the future.The concentration of total nitrogen in the raw overlying water of Shiwulihe River was31.92 mg·L^-1,the concentration of ammonia was 29.01 mg·L^-1,the concentration of nitrate nitrogen was 0.49 mg·L^-1,the nitrite nitrogen concentration was 0.02 mg L^-1.After artificial aeration treatment,the total nitrogen concentration in the overlying water is 0.88mg·L^-1,ammonia concentration is 0.61 mg·L^-1,nitrate nitrogen concentration is 0.10mg·L^-1,nitrite nitrogen concentration is 0.11mg·L^-1.The nitrogen index meets the Class III water quality standards in the Environmental Quality Standard for Surface Water（GB3838-2002）.Indicating that this aeration method has a good effect on the removal of nitrogen in the overlying water.Before creating a model using near-infrared spectroscopy in order to obtain better modeling results,to ensure a robust prediction model,it requires the use of pretreatment method to remove the interference information.In this experiment,the standard normal transform,the multi-scattering correction and the wavelet transform were used to pretreat the near-infrared spectral data respectively.Then,BP neural network（BP）,extreme learning machine（ELM）and support vector machine（SVM）were used to build the quantitative analysis model of total nitrogen,ammonia nitrogen,nitrate nitrogen and nitrite nitrogen.The results of the total nitrogen quantitative analysis model show that the best model effect can be obtained by using the SVM method after using the WT,and the r_c is0.9741,and the r_p is 0.9649,the RMESCV is 6.5382 and the RMSEP is 6.8503.The results of ammonia nitrogen quantitative analysis show that the optimal model effect can be obtained by using the method of ELM by using WT before modeling for ammonia nitrogen.The r_c is 0.9831,the r_p is 0.9679,The RMSECV is 5.6568 and the RMSEP is 5.6619.The results of nitrate nitrogen quantitative analysis show that the optimal model effect can be obtained by using the method of ELM by using WT before modeling for nitrate nitrogen.The r_c is 0.9819,the r_p is 0.9769,The RMSECV is 1.7711 and the RMSEP is 2.3163.The results of the nitrite nitrogen quantitative analysis model show that the best model effect can be obtained by using the SVM method after using the WT,and the r_c is 0.9878,and the r_p is 0.9789,the RMESCV is 0.7704 and the RMSEP is 1.2650.By the above method,the total nitrogen,ammonia nitrogen,nitrate nitrogen and nitrite nitrogen in overlying water can be well predicted.Experimental studies have shown that the use of artificial aeration method can effectively remove nitrogen from contaminated rivers.Using near infrared spectroscopy combined with stoichiometric method to establish the quantitative analysis model can quickly and non-destructively reflect the total Nitrogen,ammonia nitrogen,nitrate nitrogen,nitrite nitrogen concentration.This provides a very important theoretical basis for the rapid quantitative analysis of these substances in practical application.