Analysis Of Water Quality And Intracellular Polymers In Denitrifying Dephosphatation Process By Near Infrared Spectrum Quantitative Analysis Model

The excessive amounts of nitrogen and phosphorus in wastewater will probably lead to eutrophication,and the key measure to control eutrophication is to control the content of phosphorus,so how to remove phosphorusefficiently is still a hotpot in the field of water pollution prevention and cure.The traditional phosphorus removal process,which absorbs phosphorus excessively in the aerobic stage and removes phosphorus through the discharge of phosphorus-rich sludge,has many drawbacks,such as the different SRT of nitrifying bacteria,the competition for carbon source and the high energy consumption,though it is proved to have the advantages of good treatment result.So a new process which can avoid above-mentioned problems is urgently called for.Denitrifying dephosphatation can remove nitrogen and phosphorus simultaneously so that the same carbon source can contribute to two processes through the alternate operation of anaerobic and anoxic conditions,which releases phosphorus largely in the anaerobic stage and absorbs phosphorus excessively using nitrate as the electron acceptor.It shows that monitoring denitrifying dephosphatationfast and timely can guarantee the stable operation ofthe process.In this article,the water quality index and the measuring techniques of substances in sludge cell based on denitrifying dephosphatation were studied,and the main conclusions are as follows.(1)The experiment using the SBR reactor to enrich the PAOs,after through anaerobic/aerobic cultivation stage,when the influent concentration of phosphorus is 7.0~7.8 mg/L,the concentration of COD 260-320 mg/L,sludge age control in 12-15 days,the effluent COD concentration is less than 30 mg/L,the total phosphorus concentration is less than 0.5 mg/L,COD removal rate reached more than 90%,the phosphorus removal rate reached more than 95%.(2)When the PAOs are enriched by anaerobic/anoxic way to cultivate DPB,when the influent concentration of phosphorus is 7.0~7.8 mg/L,COD concentration control between the 160~180mg/L,anoxic nitrate nitrogen dosage control between the 18~20mg/L and sludge age control in 12 ~15d,effluent COD concentration is less than 30 mg/L,total phosphorus concentration less than 0.5 mg/L,COD removal rate reached more than 90%,phosphorus removal rate reached more than 95%.(3)The experiment respectively in water total phosphorus,nitrate nitrogen,COD and intracellular PHB,poly-P and Gly each established three quantitative analysis models of interval partial least squares(iPLS),BP neural network,the ultimate learning machine(ELM).In order to eliminate the random noises,extracting spectral information to reduce interference,improve the model's accuracy,the original spectra were pretreated by two order derivative,multiple scatter correction,wavelet denoising and principal component analysis.According to the model of correlation coefficient and root mean square error of analysis,experiment,it is found that in the process of water modeling using wavelet denoising based on principal component analysis for the optimal spectral pre processing method,while in the process of modeling the intracellular material,a more accurate prediction model can be obtained by using multiple scatter correction and principal component analysis.(4)Modelling a substance in water samples,the model built by using wavelet de-noising as the pre-process method and iPLS as modelling method will bring the optimal effect.TP correlation coefficient correction and predictive correlation coefficients were 0.9675 and 0.9294 separately;RMSECV and prediction of the root mean square error were 0.0196 and 0.0235 respectively.The correlation coefficient of nitrate nitrogen correction and predictive correlation coefficients were 0.9053 and 0.9014 separately,and RMSECV and RMSEP were 0.0295 and 0.0298 respectively.The correlation coefficient COD correction and predictive correlation coefficients were 0.8092 and 0.7848 separately.RMSECV and RMSEP were 0.1423 and 0.1568 respectively.(5)Modelling intracellular substances,the model built by using multiple scatter correction as pre-process methodand ELM as modelling method will getthe optimal effect.The correlation coefficient PHB correction and predictive correlation coefficients were 0.9835 and 0.9683 separately,and RMSECV and RMSEP were 0.0541,0.0668 respectively.Poly-P correlation coefficient correction and predictive correlation coefficients were 0.9499 and 0.9288 separately,and RMSECV and RMSEP were 0.0579 and 0.0776 respectively.The correlation coefficient Gly correction and predictive correlation coefficients were 0.9589 and 0.9488 separately,and RMSECV and RMSEP were 0.0489 and 0.0501 respectively.The results showed that modelling the NIR data of water samples and intracellular polymers by the three chemometrics methods of iPLS?BP neural network and ELM learning machine could characterize the change ofwater quality and the concentration and content of intracellular polymers in the denitrifying dephosphatation process fast and accurately,which provide real time monitoring with theoretical basis,and has important value on the operation of monitoring technology.