Study On Determination Of Pollutants In Water Based On SERS Technology And GWO-SVR Algorithm

Water is the source of life,quality water resources is the basic guarantee of human life and the survival of plants and animals.However,in recent years,water pollution incidents occur frequently,which seriously affects the social environment and even endangers human life.Among them,the abuse of antibiotics,pesticides and fertilizers is particularly serious.Although the content of these pollutants in water is not high,they pose a great threat to human health.Traditional water quality detection methods are usually complicated in operation and time-consuming in experimental process,which cannot meet the requirements of rapid and accurate detection.Therefore,it is of great significance to find a rapid and accurate detection method for trace amounts of such pollutants.Based on this,this paper studies the detection of contaminants in water by surface-enhanced Raman spectroscopy,and combines this technology with stoichiometry method to realize the prediction of solution concentration.First of all,the research background and significance of this topic were introduced,and some common detection methods of antibiotics,total nitrogen(TN),total phosphorus(TP)and spectroscopic technology for water quality detection were investigated.The basic principle of surface-enhanced Raman spectroscopy and the preparation method of substrate are described.Secondly,the method and principle of preparing nano-sized silver sol were introduced.Nano-sized silver particles with different volumes and diameters were prepared by controlling the addition amount of sodium citrate and heating time.And through the analysis of UV absorption spectrum,background signal and time stability to explore the properties of nano silver sol,to determine the best preparation scheme.Then,the SERS spectrum of the solution to be tested was measured and the characteristic peaks were assigned.The optimal mixing ratio and mixing time of the solution to be tested and the silver sol were explored,and the diluted sample solution was divided into training set and test set.Background deduction and spectral smoothing were performed on the original spectral data.The characteristic peak-to-peak strength of the sample in the test set was taken as the input value,and the corresponding solution concentration was taken as the output.The Grey Wolf Optimization Algorithm(GWO)was used to optimize support vector regression(SVR)to establish the GWO-SVR regression prediction model,and compared with ABC-SVR and PSO-BP model,it shows the superiority of the algorithm.Finally,the simultaneous detection of TN and TP in the mixed solution was studied.By analyzing the characteristic peaks and properties of the mixed solution,the characteristic peaks of TN and TP were extracted respectively to establish a regression prediction model,and the correlation coefficient and mean square error were taken as the evaluation indexes of the model for analysis.