| The situation of air quality in China is still severe,and air quality has become a critical factor for the ruling ability of various regionss.Therefore,the establishment of a long-term effective air quality warning system has become a hot issue.Under the traditional way,the air quality early warning research only focuses on six main pollutants in the air quality,and ignores the influence of influencing factors such as meteorological vehicles on the air quality.In order to solve these problems,this paper proposes an air quality early warning method based on Grey Support Vector machine,and establishes a new air quality early warning system using Java language and WebGIS technology.Based on the resource and environmental carrying capacity of Shaanxi Province,this paper combines the air quality status of Xi’an in recent years,based on the annual average data of Xi’an air quality impact factors from 2013 to 2017,using fuzzy mathematics comprehensive evaluation method for PM2.5,PM10,SO2,CO,NO2,O3 were evaluated to analyze the main pollutants of air quality in Xi’an.Through the analysis of air condition by grey relation analysis,this paper comprehensively evaluates six main pollutants,six meteorological factors and the amount of motor vehicles in the atmosphere.According to the grey relational analysis,the relational degree of each air quality influencing factor and air quality index is obtained,and the relational degree of each influencing factor is ranked.In this study,six major influencing factors of air quality,fine particulate matter,inhalable particle,nitrogen dioxide,average temperature,carbon monoxide,and sunshine time were finally selected to conduct grey theory GM（1,1）modeling.By modifying the resolution of the grey theory model and adding the preset migration parameters,the improved model is compared with the original model.It is found that the precision of the improved grey theory prediction model has been significantly improved.The improved model was used to model the six air quality influencing factors with the highest evaluation of correlation degree,and the annual mean values of the six influencing factors in the next three years were obtained.Using the sample data to supervise the support vector machine（SVM）model,the cross-validation found that the adjusted SVM model can achieve more than 90%prediction accuracy,indicating that the SVM model has high credibility,combined with the SVM model.Classify the air quality of Xi’an in the next three years,and obtain the annual air quality grade of Xi’an from 2018 to2020.Use the 2018 national monitoring data of Xi’an to test the verification results.Finally,through the development of the front and back end of the Java server and the development of the technology of the WebGIS,the air warning system is built,the data management is realized,and the early warning results are visualized. |
