Research On XRF Intelligent Inversion Algorithm For Heavy Metal Element Detection In Soil

As a country with highly intensive agricultural production,my country is facing increasingly serious soil heavy metal pollution problems.Research on efficient,accurate and convenient soil heavy metal detection methods is of great significance for understanding soil pollution and carrying out pollution prevention and control.Because X-ray fluorescence spectrometry（XRF）technology has the advantages of fast,accurate and non-destructive,it has been widely used in the detection of element content.In this paper,relying on the National Natural Science Foundation of China,the XRF analysis method for the detection of heavy metal elements in soil is carried out to solve the problems of complex nonlinear relationship between calibration curves caused by poor original spectral signal-to-noise ratio,spectral line overlap and soil matrix effect in XRF analysis.Research on Algorithms.The relevant research contents and results are as follows:（1）The preprocessing method of XRF spectrum is studied.The noise is eliminated by moving average,smooth filtering and wavelet transform,and baseline correction is performed by iterative discrete wavelet transform（IDWT）and asymmetric weighted penalized least squares（arPLS）,so that Cr,Cu,Zn,As,Pb elements are calibrated The coefficient of determination of the curve was improved from 0.965,0.979,0.971,0.794,0.915 to 0.979,0.987,0.981,0.828,0.953.（2）The intelligent inversion model of soil heavy metal elements is studied.In view of the spectral line overlap interference and matrix effect existing in soil XRF spectra,the competitive adaptive reweighting algorithm（CARS）was used to select sensitive spectral lines for different heavy metal elements.And adopt the support vector machine regression（SVR）model optimized by particle swarm algorithm（PSO）to complete the inversion of the content.The coefficient of determination of the training set and the test set is above0.99 and 0.89.Compared with the support vector machine regression and partial least squares regression models,the prediction accuracy has been improved.（3）Research to use Monte Carlo method to improve the accuracy of model prediction.Since 57 standard samples were not enough for intelligent algorithm analysis,the content information of 214 soil standard samples was obtained through the national standard material resource sharing platform,and Monte Carlo simulation was used to generate spectra and normalize them.Based on the leave-one-out cross-validation method,it is proved that the determination coefficients of Cr,Ni,Cu,and Zn elements are increased by 0.0036,0.0065,0.0117,and 0.0105,respectively.（4）To study the pollution of soil heavy metals in the Taihu Lake Tourist Resort in Huzhou.According to the sampling plan,30 soil samples were collected and processed,analyzed,submitted for inspection,and verified.The heavy metal contents of some samples submitted for inspection were compared with those predicted by the inversion algorithm.Kriging spatial interpolation analysis of Pb element content,so as to understand the pollution status of soil heavy metals in the study area.To sum up,this study completed XRF spectral preprocessing based on methods such as wavelet transform and arPLS,used CARS algorithm to automatically filter characteristic spectral lines,and used PSO-optimized SVR to establish an inversion model and Monte Carlo simulation to improve the accuracy of the model.,which provides innovative methods and technical support for the detection of heavy metal elements in soil.