Chen Gang, Chen Shen, Li Dong, Chen Cai
School of Management, Guizhou University, Guiyang, 550025, Guizhou, China.
Digital Transformation and Governance Collaborative Innovation Laboratory, Guizhou University, Guiyang, 550025, Guizhou, China.
Sci Rep. 2025 Jan 29;15(1):3685. doi: 10.1038/s41598-025-88086-1.
Air pollution is a critical global environmental issue, further exacerbated by rapid industrialization and urbanization. Accurate prediction of air pollutant concentrations is essential for effective pollution prevention and control measures. The complex nature of pollutant data is influenced by fluctuating meteorological conditions, diverse pollution sources, and propagation processes, underscores the crucial importance of the spatial and temporal feature extraction for accurately predicting air pollutant concentrations. To address the challenges of data redundancy and diminished long-term prediction accuracy observed in previous studies, this paper presents an innovative approach to predict air pollutant concentrations leveraging advanced data analysis and deep learning methods. The proposed approach, termed KSC-ConvLSTM, integrates the k-nearest neighbors (KNN) algorithm, spatio-temporal attention (STA) mechanism, the residual block, and convolutional long short-term memory (ConvLSTM) neural network. The KNN algorithm adaptively selects highly correlated neighboring domains, while the residual block, enhanced with the STA mechanism, extracts spatial features from the input data. ConvLSTM further processes the output from STA-ConvNet to capture high-dimensional temporal and spatial features. The effectiveness of the KSC-ConvLSTM approach was validated through predictions of PM concentrations in Beijing and its surrounding urban agglomeration. The experimental results indicate that the KSC-ConvLSTM approach outperforms benchmark approaches in single-step, multi-step, and trend prediction. It demonstrates superior fitting accuracy and predictive performance. Quantitatively, the proposed KSC-ConvLSTM approach reduces the root mean square error (RMSE) by 4.216-8.458 for prediction averages of 1-12 h of PM in Beijing, compared with the benchmark approach. The findings show that the KSC-ConvLSTM approach shows considerable potential for predicting, preventing, and controlling air pollution.
空气污染是一个严峻的全球环境问题，快速的工业化和城市化使其进一步恶化。准确预测空气污染物浓度对于有效的污染预防和控制措施至关重要。污染物数据的复杂性质受到气象条件波动、污染源多样以及传播过程的影响，这凸显了时空特征提取对于准确预测空气污染物浓度的至关重要性。为应对先前研究中观察到的数据冗余和长期预测准确性下降的挑战，本文提出了一种利用先进数据分析和深度学习方法来预测空气污染物浓度的创新方法。所提出的方法称为KSC-ConvLSTM，它集成了k近邻（KNN）算法、时空注意力（STA）机制、残差块和卷积长短期记忆（ConvLSTM）神经网络。KNN算法自适应地选择高度相关的相邻区域，而通过STA机制增强的残差块则从输入数据中提取空间特征。ConvLSTM进一步处理STA-ConvNet的输出以捕获高维时空特征。通过对北京及其周边城市群的PM浓度进行预测，验证了KSC-ConvLSTM方法的有效性。实验结果表明，KSC-ConvLSTM方法在单步、多步和趋势预测方面优于基准方法。它展示了卓越的拟合精度和预测性能。从定量角度来看，与基准方法相比，所提出的KSC-ConvLSTM方法在北京1-12小时PM预测平均值方面将均方根误差（RMSE）降低了4.216-8.458。研究结果表明，KSC-ConvLSTM方法在空气污染预测、预防和控制方面具有巨大潜力。
