Optimization Of Air Quality Monitoring Network

The premise of effective control of air pollution is to master detailed information of air pollutants,so countries around the world have established air quality monitoring networks to monitor pollutant data.However,these monitoring networks all have the problem of unreasonable layout,resulting in a large number of monitoring stations monitoring similar pollutants,thus forming the monitoring station redundancy.This incurs high costs and greatly reduces monitoring efficiency.In this paper,two comprehensive methods are proposed to identify redundant monitoring stations in the air quality monitoring network,identify the replacement stations of these redundant monitoring stations,and verify the effectiveness of the proposed methods by checking the new optimized air quality monitoring network.This paper takes the air quality network of Shanghai as an empirical study.The data used are the monitoring data of six major pollutants(PM_2.5,PM₁₀,SO₂,NO₂,O₃,CO)at 10 national control stations from 2014 to 2018.Air pollutant data in Shanghai are collected and the original data were preprocessed.Through the study of the processed pollutant data,it is found that the concentration of various pollutants among monitoring stations in Shanghai presents similar spatial and temporal distribution characteristics,that is,there may be redundant stations in the network.Secondly,after confirming the existence of redundant monitoring stations in Shanghai air quality monitoring network,a follow-up study is carried out.The first proposed comprehensive method for air quality monitoring network optimization uses:（i）Correlation analysis and principal component-multiple linear analysis to determine the correlation of pollution information monitored by other stations and to find redundant stations and their replacement stations;（ii）Cluster analysis and correspondence analysis to quantitatively validate alternatives to redundant monitoring stations;（iii）Support vector machine regression is used to verify the new optimized air quality monitoring network.Finally,the effectiveness of the optimized air quality monitoring network is verified.Then the second comprehensive method for air quality monitoring network optimization is proposed,which uses（i）correlation analysis and stepwise regression analysis to identify redundant stations and their replacement stations;（ii）Cluster analysis and correspondence analysis to validate alternatives to redundant monitoring stations;（iii）BP neural network regression to verify the optimized air quality monitoring network.Through the above comprehensive methods,four redundant stations（Xuhui,Zhangjiang,Shiwuchang and Pudong New Area）are determined,and the alternatives of each redundant station are found.And the new air quality monitoring network is verified to prove the rationality and effectiveness of the proposed methods.Finally,the two comprehensive methods are compared and analyzed,and the identification of redundant stations and the final prediction and verification methods are the biggest differences between the two comprehensive methods.The methods of identifying redundant stations（Principle component-multivariate linear analysis,stepwise regression analysis）effectively compensate for the defects of the methods used in previous research.And the stepwise regression method further optimizes the identification of redundant stations by principle component analysis.The final verification methods（Support Vector Regression,BP neural network prediction）verify the rationality and effectiveness of the final optimization scheme,which are also lacking in the previous research.The research results show that although the methods selected are not the same,the conclusion is the same,that is,the two comprehensive methods can effectively identify the redundant monitoring stations in the air quality monitoring network,so as to optimize the air quality monitoring network,and improve the monitoring efficiency.