Analysis Of Spatial And Temporal Variation Of PM_2.5 And Meteorological Influencing Factors In Anhui Province

In this paper,16 cities in Anhui Province were taken as the research objects.Based on air quality data and meteorological observation data from 2015 to 2019,the spatial and temporal distribution characteristics of PM_2.5concentration were discussed and analyzed from different scales by combining with MINITAB,and the correlation between air pollution factors and meteorological factors and PM_2.5concentration in Anhui Province was discussed with the help of SPSS software.Principal component analysis was used to select the main impact factors,and PM_2.5concentration prediction was conducted based on MATLAB platform.BP neural network model and cubic exponential smoothing model were established respectively.This will be of positive significance to the control of PM_2.5pollution in the future.The main research results of this paper are as follows:?1?Air quality index in Anhui province in 2015-2019,the air quality is best,Huangshan in Anhui province,the optimal air ranks first,overall,most of 16 cities in Anhui province is a good air quality,from 2015 to 2016,the air quality has improved,but the air quality in 2017 fall,many cities appear light pollution,even moderate pollution phenomenon,whole is between 33.3%?91.7%.In 16 cities of Anhui Province,the concentrations of PM_2.5,PM₁₀,SO₂and CO ranged from 35 to 65?g/m³,71 to 81?g/m³,9.56 to 22.13?g/m³and 0.75 to 0.97mg/m³.The concentration of NO₂fluctuated in 5years,mostly distributed in 25?35?g/m³.O₃concentration showed an increasing trend and decreased in 2019,mostly distributed in the range of 75?115?g/m³,annual mean range of 69.13?101.4?g/m³,standard deviation range of 7.79?16.02?g/m³,O₃concentration difference in each city is gradually decreasing.?2?The overall concentration of PM_2.5in 16 cities in Anhui was as follows:>in the north of Anhui,>in the middle of Anhui,and south of Anhui.The variation ranges were20?123ug/m³,17?102ug/m³,and 10?103ug/m³,respectively.On a seasonal scale,the concentration of PM _2.5in 16 cities of Anhui province was greater in winter than in spring,greater in spring than in autumn,and greater in autumn than in summer.On the monthly scale:generally speaking,the variation shows the v-shaped characteristic of"low weekly high".PM_2.5concentration in most cities peaks in December and January,and then decreases in fluctuation.When the PM_2.5concentration reaches the annual concentration trough in July and August,then gradually increases with the coming of autumn and winter.On a daily scale,the variation of PM_2.5concentration presents a distinct bimodal trend,with a small fluctuation in summer and a large fluctuation in winter.Therefore,the fluctuation of peak and valley value varies with different seasons.?3?From the perspective of multiple regression,PM_2.5concentration is strongly positively correlated with PM₁₀and CO,PM_2.5is positively correlated with SO₂and NO₂,and negatively correlated with O₃,and Pearson value reaches 0.741.From the point of meteorological factors,PM_2.5concentrations in change,and negatively correlated with mostly single meteorological factors,such as precipitation,average temperature,average relative humidity,sunshine time,maximum wind speed,etc.,but positively correlated with the average pressure,but affected by seasonal factors,rainfall in winter and PM_2.5concentrations presents a strong negative correlation,the average pressure stronger correlation in the summer.In this paper,the concentration of PM_2.5in Anhui province was analyzed by principal component analysis,and it was found that PM_2.5was extremely correlated with PM₁₀,CO,average atmospheric pressure,average relative concentration,maximum wind speed and SO₂,realizing feature extraction.?4?The prediction model was built:BP neural network model and cubic exponential smoothing model were used for prediction.The results showed that BP neural network had less error than the traditional exponential smoothing method,but the training time was longer.Therefore,the three-time exponential smoothing neural network combination model was built to predict PM_2.5concentration,and the fitting degree was 86%.The uncertainty analysis of the three-time exponential smoothing neural network model was carried out in combination with MAPE and Kappa,and the output result was highly consistent with the actual value.At the same time,a PM_2.5Concentration Intelligent Data Analysis and Prediction System is designed to realize PM_2.5concentration observation data,correlation analysis,model prediction and other functions.