The Influencing Factors And Prediction Analysis Of PM2.5 In Hangzhou

With the rapid development of China’s economy,more and more cities have air pollution,which has become a widespread concern.Among the air pollutants,PM2.5as the main pollutant,PM2.5 concentration has also become an important indicator to measure the air quality of a city.Therefore,we hope to predict the concentration of PM2.5 effectively and scientifically.Taking Hangzhou as the research object,this paper selected the pollutant data of Hangzhou and its surrounding cities in recent years for analysis.Firstly,the correlation between PM2.5 concentration and the average temperature,humidity,average wind speed,and atmospheric pressure of Hangzhou was analyzed.It was found that PM2.5concentration and average temperature were negatively correlated,positively correlated with atmospheric pressure,weakly correlated with average humidity and wind speed.Then,the correlation analysis of PM2.5 concentration and the other five air pollutants is carried out.The results showed that PM2.5 concentration and PM10concentration,sulfur dioxide（SO₂）,carbon monoxide（CO）,nitrogen dioxide（NO₂）had strong correlation,so it is necessary to reduce the concentration of these pollutants in priority.Then the air quality index（AQI）and the air pollutant index were regressed,and PM2.5 was found to have the greatest impact on the air quality index.Finally,the PM2.5 concentration correlation in Hangzhou,Jiaxing,Huzhou,and Shaoxing was analyzed,and the regional air pollution problems in Hangzhou were explored.It was found that the air quality in several districts near the West Lake District was the best,while in Xiaoshan District,Yuhang District and Lin’an District were relatively poor.Then,a time series model was established to predict and analyze PM2.5concentration.After a series of data processing,ARIMA（1,1,2）was used to fit the data.After the residual test of the model,it was found that the residual was a white noise sequence,which indicated that the model fitting was successful.In addition,the PM2.5concentration in the last six days of the modeling data was predicted.The results showed that the prediction accuracy of the first four days was high,and the prediction accuracy of the next two days started to get worse,which indicated that the time series model can only predict the PM2.5 concentration in a short term,and in the medium term,there were some deficiencies in the long-term forecast.Because BP neural network has strong nonlinear processing ability,this paper used BP neural network modeling to predict the long-term PM2.5 concentration,used1092 data from 2016 to 2018 to train the neural network model,after repeated parameter debugging,it was determined that the input layer neuron was 7,the hidden layer neuron was 5,and the output layer neuron was 1,The seven input variables of the model were used to analyze and predict the 17 data after the training data.With comparing with the actual value,the overall relative error between the predicted result and the actual result was 8.96%.Then,through the promotion,the PM2.5concentration in the first half of 2019 was predicted,the overall relative error was11.12%,and the proportion of the predicted acceptable days was 87.57%,which showed that BP neural network had a good effect on medium and long term prediction,and provided an effective method for PM2.5 concentration research.At the end of the paper,some suggestions are put forward to reduce the air quality index of Hangzhou,such as reducing the emission of sulfur dioxide from the source,desulfurizing the iron and steel,cement and chemical industries,reducing the emission of nitrogen dioxide and carbon monoxide,promoting the use of new energy vehicles,promoting the citizens to reduce the use intensity of motor vehicles,and getting out by carpooling or public transportation or sharing single cars.At the same time,we need to train environmental management professionals to jointly manage air pollutants with surrounding cities.