Prediction Of PM2.5 Concentration In Wuhan Based On Ssa Combined Model

Since this century,the problem of air pollution in China has been more and more terrible,which has seriously affected people’s normal activities and work,and has a negative impact on the health of ordinary people.The problem of air pollution is a key issue of great concern to our country,and it is also related to the life of every ordinary person.Among the air pollutants,the most influential one is the fine particles in the air(commonly known as PM2.5).This kind of pollutants mainly come from the pollution of modern industry,the harmful exhaust gas emitted by motor vehicles,etc.it has a threat to human health,and directly affects the respiratory system,cardiovascular system,etc.Therefore,it is of great significance to monitor the concentration of PM 2.5 in the atmosphere,and to study effectively and scientifically the early warning and prediction according to the monitoring results.This paper collects the historical data of air quality in Wuhan from 2014 to 2019,makes descriptive statistical analysis on it,excavates the characteristics of air quality change in Wuhan,and establishes a combined prediction model for the daily concentration data of PM 2.5.First of all,the singular spectrum analysis(SSA)method in signal processing is applied to the decomposition of time series data.The original time series data is decomposed into the main components with trend and seasonality and the noise components with large fluctuation amplitude.Using the idea of "divide and conquer",a prediction model is established for the two components respectively,and ARIMA model,SVR model and LSTM model are used,the prediction results are combined into the final prediction results.In order to test the prediction effect of the combined model used in this paper,it is compared with the prediction result of a single basic prediction model.The final results show that the prediction accuracy of the model after SSA decomposition is higher than that of the basic model alone.In addition,in the combined model constructed,the combination model obtained by using SSA to decompose the sequence,LSTM to predict the main components and ARIMA to predict the noise components has better prediction effect than other models It has certain advantages.In this paper,we combine the singular spectrum analysis algorithm with the ARIMA model to construct the combined model and compare it with other single models and combined models.The feature of this paper is mainly reflected in the application of singular spectrum analysis,a filtering method in signal processing,to one-dimensional time series data.The time series data is processed by singular spectrum decomposition,and the noise components representing the uncertainty are extracted,so as to effectively weaken the influence of uncertainty factors on the time series model.The combined prediction model based on this can effectively predict the PM2.5 daily average concentration compared with the single time series prediction model,with higher accuracy and better prediction effect.