Prediction of PM2.5 concentration in Xi 'an city based on BP neural network

In view of the deficiency of the traditional PM2.5 concentration prediction model, this paper proposes a PM2.5-hour concentration multi-step prediction model based on BP neural network, combining pm2.5 influencing factors with its historical concentration data as input to the prediction model. Firstly, according to the air quality detection data from December 1, 2019 to November 30, 2020 in Xi'an, the correlation analysis of the effects of PM10, SO2, NO2, CO and O3 air pollutants on PM2.5 concentration was carried out. Then, based on the analysis results of relevant influencing factors, the current forecast values of PM10, SO2, NO2, CO, O3 and the historical concentration of the k-order of PM2.5 were determined as input values of the neural network, and the optimal value of the historical concentration order k was determined by trial. The BP neural network model is used to simulate and predict pm2.5 concentrations in different times in Xi'an, and the final results show that the prediction model can obtain accurate prediction values in each time.