PM_2.5 Concentration And Visibility Forecast In Beijing Tianjin And Hebei Region Based On Machine Learning

In recent years,with the development of the economy and population,air pollution in our country has become more and more serious.Heavy pollution has appeared in many regions,and it has a long time and a large scope.Especially in the Beijing-Tianjin-Hebei and surrounding areas,the pollution problem is the most serious.Therefore,it is of great significance to study the efficient and real-time air quality forecasting system,which can not only provide guidance for people to travel,but also provide technical support for functional departments to prevent and control polluted weather.The air quality forecast model is a hot research topic at present,but due to the development of technology and physical condition,traditional statistical forecast and foreign numerical forecast model can not meet the operational needs of air pollution forecast and early warning.In view of the above situation,this paper proposes a PM_2.5 concentration forecast model based on hierarchical sparse representation,and proposes a visibility forecast model combined with environmental factors based on hierarchical sparse representation.Firstly,this paper selects meteorological stations with altitudes below 1km in the Beijing Tianjin and Hebei region.Then it analyzes the temporal distribution characteristics of PM_2.5concentration and visibility.and it analyzes their correlation coefficient of Pearson with meteorological factors.The results show that the pollution status of Beijing Tianjin and Hebei region is relatively serious,and the proportion of the pollution of PM_2.5 is relatively large,mainly in January,February,March,November and December.Secondly,this paper takes Beijing and other stations as examples,establish a forecast model of PM_2.5 concentration based on the relevant influence factors of PM_2.5 combined with the fuzzy C-means clustering?FCM?algorithm and sparse representation algorithm,and predicts the PM_2.5 concentration of the main stations of Beijing Tianjin and Hebei region.The forecast results are compared with the result of model and analyzed in two aspects.The results show that the model can shorten the forecast time effectively,improve the accuracy of PM_2.5 concentration.Finally,this paper uses the correlation analysis of visibility to select the factors of visibility forecast,and use the fuzzy C-means clustering?FCM?algorithm and the hierarchical sparse representation algorithm to establish the visibility forecast model.This paper predict the visibility of the main stations in Beijing Tianjin and Hebei region in winter,and compared with the existing models and BP neural network methods.The forecast results are analyzed in many aspects.The results show that the forecasting factor combined with the concentration of pollutants can effectively improve the forecast accuracy of low-visibility weather.It can be seen that the PM_2.5 concentration and visibility prediction model established by the fuzzy C-means clustering?FCM?algorithm and the hierarchical sparse representation algorithm can predict effectively.It verifies the potential of this method in PM_2.5concentration and visibility prediction.The model based on hierarchical sparse representation is simple and easy to expand,which improves the precision of PM_2.5 concentration and visibility forecast,and is convenient for other meteorological analysis.