Air Quality Analysis And Forecasting In Harbin Based On The Artificial Neural Network

Air pollution not only leads to the deterioration of environmental air quality,but also affects human health and the development of regional economy.The issue has attracted extensive attention from the government,the public and relevant scholars.In recent years,with the development of industrialization and urbanization,the consumption of massive fossil energy has made air pollution increasingly serious.As a typical heavy industry city,Harbin is faced with severe air pollution,especially during haze days in winter.This study selected Harbin as the main research area.The relevant data of air quality index（AQI）,concentration of six major air pollutants(PM_2.5,PM₁₀,SO₂,NO₂,CO and O₃)and four major meteorological factors（precipitation,average temperature,average wind speed and average relative humidity）in Harbin city from January 1,2016 to December 31,2020 were collected.Combined with the pre-processed data and based on statistical analysis theory and artificial neural network,the air quality of Harbin was studied as follows:（1）Explore the law of air quality changes over time.Based on air quality data of Harbin,the empirical analysis is made from three aspects:annual,monthly and heating period.The purpose is to explore the temporal variation of air quality.The results showed that the air quality in Harbin has not been effectively improved in the past five years,and the air quality of Harbin presents an obvious seasonal variation:the air pollution is serious in winter,light in spring and autumn,and the best in summer.In addition,compared with the non-heating period,the air quality in the heating period of Harbin is worse.（2）Analyze the influencing factors of air quality in Harbin.Through Pearson correlation coefficient,this study explored the correlation between AQI and six major air pollutants.The results showed that air quality has an extremely strong positive correlation with PM2.5 and PM10,a strong positive correlation with SO₂,CO₂,CO,and a negatively weak correlation with O₃.Additionally,air quality has a negatively weak correlated with meteorological factors,and the air pollution will get improved when the temperature is higher,the rain is more,the wind is stronger and the air relative humidity is higher.（3）Establish an artificial neural network prediction model for air quality in Harbin.The concentration of PM_2.5,PM₁₀,SO₂,NO₂,CO and AQI,which are highly correlated with air quality according to previous research results,were used as data sets to establish Back Propagation（BP）neural network,Long Short-Term Memory（LSTM）neural network and PSO-LSTM neural network prediction model optimized by Particle Swarm Optimization.In this study,AQI,PM2.5,PM10,SO2,NO2 and CO were used as the input data of the model,and AQI was used as the output data of the model.The mean absolute percentage error（MAPE）,mean absolute error（MAE）,root mean square error（RMSE）and determinable coefficient（R²）were used as the evaluation indexes of the model.The results show that all those models can effectively predict air quality,and the prediction effect of PSO-LSTM neural network model is better than that of BP neural network model and LSTM neural network model.Accurate air quality analysis and prediction is conducive to the comprehensive control of urban air quality.The analysis of the time changes regulation and the influencing factors of air quality have important instructive significance for the improvement of urban air quality and the control of air pollutants.The prediction of air quality can not only help the public to grasp the urban air quality information in advance and provide reference for outdoor activities,but also provide effective suggestions and help government departments to formulate policies related to air quality improvement.Therefore,air quality analysis and prediction have important theoretical significance and practical value for the improvement of urban air quality and the improvement of life quality for the public.