Application Of Air Quality Prediction In The Yangtze River Delta Based On LSTM Model

In the past several years,air pollution has gradually become a severe factor affecting people’s lives.Due to the sharp growth of industrialization and the considerable cluster of urban population in the Yangtze River Delta region,it has had a profound impact on the regional atmospheric environment.Therefore,the prediction of the air quality in the Yangtze River Delta that provides a powerful reference for our daily life and relevant government departments to formulate environmental protection policies.The Air Quality Index（AQI）is based on the environmental air quality standards and the impact of various pollutants on human health,ecology and the environment,reflecting the pollution of the six factors(SO_2,O₃,PM2.5,PM10,NO₂,and CO).The pollution is closely related to the exhaust gas emitted by the burning of fossil fuels or automobiles.This paper takes cities in the Yangtze River Delta region as the target area,uses Python crawlers to obtain daily monitoring data of the air quality index of cities in the region from January 2015 to December 2019,and establishes a model to predict.The efficiency and accuracy of traditional regression prediction models are often unsatisfactory.However,the recurrent neural network model can take advantage of the medium and long-term dependence on information of time series data to perform relevant predictive analysis on such data.However,there are problems in the recurrent neural network model that the gradient disappears and the parameter information is overloaded.Consequently,this paper uses LSTM（Long Short Term Memory Neural Network Model）,that introduces a threshold system on the basis of RNN.The parameters of the LSTM should be determined by some ways.This paper introduces the MFO algorithm to look for the parameter space.The MFO algorithm proposes a new type of swarm intelligence optimization algorithm in simulating the spiral flight path of the moth based on the navigation mechanism of the moth during flight.Finally,through GRA and K-means,the primary factor of air pollution and the cities with similar air pollution in the Yangtze River Delta can be acknowledged.Compared with the prediction of SVM and BP,using RMSE and MAE as model evaluation indicators,the precision of LSTM model prediction have been greatly improved.Combined with GRA to get the primary air pollution factors of each city and K-means to find cities with similar air pollution situation,it has important reference significance for people’s daily life and the designation of environmental protection policies by relevant government departments.