PM2.5 Concentration Prediction Based On Time Series Decomposition

In recent years,large and medium-sized cities in China are frequently covered in haze,which poses a great threat to the atmospheric environment quality and people's health.Fine particulate matter(PM2.5)is one of the important components of haze.Severe PM2.5 pollution not only increases the risk of acute respiratory diseases,asthma and lung cancer,but also shortens human life.PM2.5 can directly or indirectly affect the climate and change the human living environmet.Therefore,the accurate prediction of PM2.5 in the future can be realized.On the one hand,it can remind people to take preventive measures in time so that won't affect their health.On the other hand,it can help the environmental protection department to improve the monitoring of PM2.5,so as to better control the haze.The main purpose of this paper is to predict the average daily concentration of PM2.5 in Lanzhou city,so a new hybrid model based on two time series decomposition methods is proposed.In order to verify the effectiveness and adaptability of the hybrid model proposed in this paper,Nanjing city,which is completely different from Lanzhou city in geographical location and atmospheric environment,was selected for model verification.The results show that the hybrid model proposed in this paper is not only more advantageous in predicting the average daily concentration of PM2.5,but also has extensive adaptability.By comparing the prediction accuracy of the mixed model under the two decomposition methods,it is found that the STL decomposition method is more accurate in the prediction of the average daily concentration of PM2.5.