Forecasting of PM2.5 Concentration in Beijing Using Hybrid Deep Learning Framework Based on Attention Mechanism

Abstract

Air pollution has become a critical factor affecting the health of human beings. Forecasting the trend of air pollutants will be of considerable help to public health, including improving early-warning systems. The article designs a novel hybrid deep learning framework FPHFA (FPHFA is the abbreviation of the title of this paper) for PM2.5 concentration forecasting is proposed, which learns spatially correlated features and long-term dependencies of time series data related to PM2.5. Owing to the complex nonlinear dynamic and spatial features of pollutant data, the FPHFFA model combines multi-channel one-dimensional convolutional neural networks, bi-directional long short-term memory neural networks, and attention mechanisms for the first time. Multi-channel 1D CNNs are applied to capture trend features between some sites and overall spatial characteristics of PM2.5 concentration, Bi LSTMs are used to learn the temporal correlation of PM2.5 concentration, and the attention mechanism is used to focus more effective information at different moments. We carried out experimental evaluations using the Beijing dataset, and the outcomes show that our proposed model can effectively handle PM2.5 concentration prediction with satisfactory accuracy. For the prediction task from 1 to 12 h, our proposed prediction model performs well. The FPHFA also achieves satisfactory results for prediction tasks from 13 to 96 h.

Hints of Earlier and Other Creation: Unsupervised Machine Learning in Financial Time-Series Analysis

“…This study's methods also invite the extension of unsupervised machine learning to forecasting tasks that supervised machine learning methods have begun to tackle. For instance, all three methods of clustering presented in [2,3] and this study can be applied to the immensely popular task of forecasting air pollution in Beijing [87][88][89]. Unsupervised machine learning holds promise for addressing similar problems in meteorology, pollution control, and other environmental sciences [90].…”

Deep-learning architecture for PM2.5 concentration prediction: A review

“…[ 132 ] 2022 Kerala, India EDPF H/M/T+24 12.96 9.28 56.73 0.44 Li et al. [ 133 ] 2022 Beijing, China FPHFA H/M/T+(1-12) 28.15 19.19 56.10 0.87 H/M/T+(13-24) 22.12 15.27 43.80 0.93 Gunasekar et al. [ 134 ] 2022 Chennai, Tamandu ARTOCL NA 0.50 0.32 - 0.69 …”

“…[ 131 ] and Li et al. [ 133 ] proposed hybrid frameworks that combined a CNN, an LSTM, and an attention mechanism, and Zhang et al. [ 131 ] considered fine-grained air pollution estimation.…”

Prediction of Pollutant Concentration Based on Spatial–Temporal Attention, ResNet and ConvLSTM

“…Numerical modeling is mainly based on meteorological principles, knowledge of atmospheric dynamics and statistics, and the construction of equations for atmospheric pollutants and meteorological data to predict short-term pollutant concentrations [11,12]. Short-term predictions of pollutants generally refer to predicting pollutant concentrations for the next 1-6 h [12].…”

“…Numerical modeling is mainly based on meteorological principles, knowledge of atmospheric dynamics and statistics, and the construction of equations for atmospheric pollutants and meteorological data to predict short-term pollutant concentrations [11,12]. Short-term predictions of pollutants generally refer to predicting pollutant concentrations for the next 1-6 h [12]. Then, according to the constructed atmospheric conditions, complex differential equations are solved by a computer to simulate the pollutants' chemical, environmental, and transportation procedures throughout the atmosphere [13].…”