Air Quality Forecast Based On Weighted KNN And Influence Factor Analysis

On February 28,2018,a press briefing on air pollution prevention and control was held by the Party Central Committee,emphasizing that the focus of work was to improve the air pollution prevention and control work in 2019 based on the achievements made in the previous year.Now the scope of pollution is expanding,the treatment of air pollution has also risen to the level of national policy.In order to better prevent and control air pollution,it is necessary to evaluate and predict air quality and find the factors that affect air quality,so as to take timely response and preventive measures.Data in this paper are obtained from China Air Quality Online monitoring platform and China Statistical Yearbook.Shenyang daily air quality in 2019 is selected for grade prediction.First of all,according to the characteristics of the data,the missing value test and data standardization are performed on the data.Neuralnet feedforward neural network model is established to predict the air quality.Six air pollutant indexes(PM2.5、PM10、SO2、CO、NO2and ozone)and three meteorological indexes(Relative humidity,temperature,precipitation)are selected in this paper.Through the establishment of neural network,the prediction accuracy of adding meteorological indicators is 98.18%,while the original index prediction accuracy is 95.45%.Secondly,for cities with high pollution degree,it is necessary to predict the air quality grade.In this paper,a weighted K-nearest Neighbor and an unweighted K-nearest neighbor model are established respectively.It is found that the prediction accuracy of the weighted K-nearest neighbor model is 97.3%,and that of the unweighted model is 95.5%.Next,this paper focuses on the impact of seasonal factors and economic factors on air quality.As for seasonal factors,this paper selects the monthly average air quality index(AQI)from 2001 to 2019 for visual analysis,and finds that the influence of season on air quality is significant.Moreover,from the fluctuation range of various pollutants and the average value of AQI,it can be seen that the air pollution degree is the lowest and the air quality is the best in summer.As for economic factors,this paper selects the air quality levels of each season of Shenyang from 2016 to 2019 and three representative indicators of economic factors(per capita GDP,per capita disposable income and per capita consumption expenditure of Shenyang).Grey correlation analysis is used to calculate the relational degree of three economic indexes and air quality indexes,the results is 0.69,0.71,0.72.Then,take the seasonal factor as the control variable,the hierarchical regression analysis is carried out for the three economic indicators and air quality levels.The final analysis results show that the higher per capita disposable income,per capita consumption expenditure and per GDP,the higher the air quality grade,all three economic factors have a negative impact on air quality.Finally,according to the results of empirical analysis,this paper puts forward some suggestions for improving the air quality in Shenyang.There are three innovations in this paper:First,meteorological indicators should be incorporated into the air quality level prediction model to make the selection of indicators more comprehensive.The force of nature is more powerful than any measure to control smog.In the process of prediction,the methods and parameters of the prediction model are optimized,and the prediction model with meteorological indexes added has higher prediction accuracy.Second,by improving the traditional algorithm of K nearest neighbor,we get the algorithm with weight to improve the prediction accuracy.The weight is added to the method,which can better assign the importance of each index.Third,this paper make a visual analysis of seasonal factors.How seasonal of different cities affect the air quality is shown objectively in this paper,and the four seasons is different of Shenyang AQI is clearly demonstrated by using data visualization.