Research On Detection Method Of Water Pollutant Content Based On Near-infrared Spectroscopy

Water resource is an indispensable natural resource.The per capita possessive quantity of water resource in our country is far lower than the world average level,so timely treatment and recycling of sewage is very important.Before the complex treatment of sewage,it is necessary to timely detect the contents of various pollutants.The detection technology based on near-infrared spectroscopy can meet the detection requirements of high efficiency and has no secondary pollution in environment,so it has become a popular detection mean and has been gradually applied in water quality detection.Based on the application of chemical oxygen demand(COD),total nitrogen and total phosphorus content detection in water,the interference signal processing method and the spectral data processing method applied in quantitative analysis were studied,to solve the problem of signal crosstalk in the miniaturized Fourier transform near-infrared spectrometer and the low accuracy of the NIR spectral prediction model.(1)An interferometric signal processing method is proposed to avoid crosstalk,which is called asynchronous equioptical-path calibration.The method,based on the non-synchronous acquisition of the sample optical interference signal and the reference laser interference signal,can avoid the crosstalk of the two interference signals from the source.The influence of the mechanical thermal noise and the change of the scanning optical path on the calibration method are analyzed.To reduce the influence of scanning optical path on asynchronous calibratal accuracy,a method of laser interference signal adjustment is designed,which is beneficial to remove the laser and further miniaturize the spectrometer.The proposed method is verified by experiments on actual water sample and configuration water sample,and the experimental results show that this method can reduce the error of prediction model by avoiding signal crosstalk.(2)To further reduce the detection error,a new near-infrared spectral transformation method,adaptive full-rank fitting integral transformation,is proposed based on the asynchronous calibration of miniature spectrometer.In this method,the number of Gaussian functions involved in the Gaussian curve fitting is determined by the discrete points of real near-infrared spectrum,and the center of the Gaussian function is determined by the wavelength position of the discrete points.The optimum bandwidth of Gaussian function is determined by correlation analysis.Based on this,the curve fitting equations are constructed,the height of the Gaussian function is determined by solving the equations,and the area integration of the Gaussian function is carried out to obtain the transformation results of the spectrum,and then the content prediction model is built.The experimental results show that this method can effectively reduce the detection error of COD,total nitrogen and total phosphorus.A common data set is also used to verify the universality of the method.