PM2.5 concentration prediction in industrial parks integrating AEC and spatio-temporal characteristics

With the purpose of realizing the fine management and control of enterprise pollution emissions and capturing response relationship between enterprise pollution emissions and pollutant concentrations in industrial parks, a PM2.5 concentration prediction model was proposed to integrate atmospheric environmental capacity (AEC) and spatial-temporal characteristics. Firstly, the daily average atmospheric self-purification capacity index (ASI) was obtained using finite volume method and combined with the daily emission data acted as AEC characteristics. At the same time, temporal characteristics at the target monitoring station and spatial characteristics including its surrounding monitoring points of PM2.5concentration were showed by wavelet analysis and Pearson correlation coefficient method. Then, in order to guarantee the fast and accurate prediction performance, the correlation characteristics of PM2.5 in training data were obtained by CNN, and BILSTM was used to fully reflect the key historical long short-term dependencies implied in time series training data. The air pollutant observation data, meteorological data and emission data of Puyang Industrial Park from 2018 to 2020 were applied to experimental verification. The results show that the CNN-BILSTM prediction model proposed in this paper improves the prediction accuracy by 10% compared with the traditional LSTM model. AEC and spatio-temporal features are well situated to improve accuracy and stability of model. The CNN-BILSTM prediction model integrating AEC and spatio-temporal features has the highest prediction accuracy during PM2.5 pollution period, which is up to 93 %. Moreover, the seasonal prediction results performed the highest prediction accuracy in autumn and winter. © 2022 Chinese Society for Environmental Sciences. All rights reserved.