Research And Implementation Of Soil Heavy Metal Content Prediction And Visualization System Based On Neural Network

With the continuous improvement of Chinese industrialization,the pollution of heavy metals in farmland in suburbs has become more and more serious.Monitoring and prevention of soil heavy metal content has become urgent.However,traditional sampling and analysis monitoring methods require huge monitoring costs,which leads to its monitoring capabilities to be limited.This paper uses artificial neural networks to establish non-linear mapping relationship,based on 1161 samples taken by the project team in six suburban areas around Wuhan to train neural network prediction model,and use limited sampling monitoring points to predict the heavy metal content of soil in places that cannot be effectively monitored due to limited monitoring costs,thereby making up for the traditional sampling and analysis methods due to limited monitoring capabilities.Lead to one-sidedness of monitoring results.At the beginning of the establishment of neural network prediction algorithm model,this paper selects four classical neural network prediction algorithm models for comparison experiments: Back Propagation Neural Network(BPNN),Wavelet Neural Network(WNN),Radial Basis Neural Network(RBFNN)and Generalized Neural Network Regression Neural Network(GRNN).Through comparative experiments,it is found that among the four neural network prediction algorithm models used in this article,the generalized regression neural network has the best prediction accuracy in the field of predicting soil heavy metal content.However,the general regression neural network still has the problem of insufficient prediction accuracy,so this paper uses the bird swarm algorithm(BSA)to optimize the smoothing factor of the general regression neural network.In the actual modeling,it was found that the bird colony algorithm has defects such as easy to fall into the local optimum,so this paper uses the parallel bird colony algorithm(PBSA)to solve the defects of the bird colony algorithm.Parallel bird swarm algorithm is used to optimize the smoothing factor of generalized regression,and the PBSA-GRNN prediction algorithm model is established.Through comparative experiments,it is proved that PBSA-GRNN has the best prediction accuracy compared with the other four neural network prediction algorithm models.This paper analyzes the actual monitoring work requirements of the first-line soil heavy metal content monitoring personnel,and designs and implements a set of soil heavy metal content prediction and visualization system.The system uses the SSM back-end framework and FastAPI back-end framework to build,and the front-end uses Vue.js front-end development framework and Element UI framework,and the data layer uses a lightweight database MySQL to store data,providing users with a spatial distribution map of sampling points and prediction points in Wuhan,and a pie chart of Nemeiro comprehensive pollution assessment in each district of Wuhan,Visualization functions such as the spatial distribution map of soil heavy metals in various districts of Wuhan;embed the above five neural network prediction algorithm models into this system to provide users with the comparison of the prediction accuracy of the five models and the prediction function of the PBSA-GRNN algorithm model.Users can upload The data set to be predicted is provided with the function of downloading predicted data for users after the system performs function calculations.The system is simple to operate,simple in interface distribution,and strong in user interaction,which can meet the work requirements of front-line soil heavy metal monitors for monitoring and prevention and control of the content of various heavy metals in the soil.