A Dynamic Grain and Semantic-FCM Model for Long-Term Multivariate PM2.5 Forecasting
DOI:
https://doi.org/10.54097/va8nxd10Keywords:
Dynamic Granular Computing, Fuzzy Cognitive Map, Long-term Multivariate Time Series Forecasting, PM2.5 ConcentrationAbstract
This paper addresses long-term multivariate time series forecasting, where feature redundancy, noise interference, and complex inter-variable relationships pose significant challenges. A ridge regression model enhanced by Dynamic Granular Computing (DGC) and Fuzzy Cognitive Map (FCM) features is proposed for long-term PM2.5 concentration prediction. The DGC module transforms raw time series into granule-level features described by statistical attributes such as mean, slope, and standard devia-tion, enabling dimensionality reduction and noise mitigation in long-horizon sequences. Since statistical granules alone cannot capture latent relationships among granule-level concepts, an FCM based on nonlinear Hebbian learning is employed to extract cognitive features. These features are combined with statistical granule features and used as inputs to the ridge regression predictor. Experimental results show that granule-based models outperform baselines using raw sequences, and that FCM-enhanced models achieve higher R² values and more stable error performance across multiple monitoring sta-tions, demonstrating the effectiveness of the proposed approach.
Downloads
References
[1] L. Zhao, Z. Li and L. Qu, “Forecasting of Beijing PM2.5 with a hybrid ARIMA model based on integrated AIC and improved GS fixed-order methods and seasonal decomposition,” Heliyon, vol. 8, no. 12, e12239, Dec. 2022.
[2] F. Xiao, M. Yang, H. Fan, G. Fan and M. A. A. Al-qaness, “An improved deep learning model for predicting daily PM2.5 concentration,” Scientific Reports, vol. 10, Art. no. 20988, Dec. 2020.
[3] Z. Liu, D. Ji and L. Wang, “PM2.5 concentration prediction based on EEMD-ALSTM,” Scientific Reports, vol. 14, Art. no. 12636, Jun. 2024.
[4] Zhao, B., Wang, J., & Yu, X. Forecasting of Beijing PM2.5 with a hybrid ARIMA model based on optimized decomposition and parameter selection. Heliyon, 2022, 8(5): e09456.
[5] Yang, H., Zhou, J., Li, Y., & Liu, W. Time-Series Forecasting of PM2.5 and PM10 Concentrations Based on the Integration of Surveillance Images. Sensors, 2025, 25(3): 8374.
[6] Rani, M., Rajasekar, P., & Sinha, P. Forecasting PM2.5 concentrations using statistical modeling for Bengaluru and Delhi regions. Environmental Monitoring and Assessment, 2023, 195(4): 456.
[7] X. Ao, J. Zhang, and L. Chen, “Analysis and prediction of PM2.5 concentration based on seasonal time series models,” Journal of Environmental Sciences, vol. 41, no. 7, pp. 2874–2883, 2021.
[8] Kim,S.,&Park,J.Long-term air pollution forecasting using Facebook Prophet model: A case study of Seoul.Atmospheric Environment, 2020, 224: 117307.
[9] Chang, L., Zhang, Z., Zhao, Y., & Chen, W. (2020). An LSTM-based aggregated model for air pollution forecasting. Environmental Science and Pollution Research, 27, 38827–38839.
[10] Qi Y, Li Q, Karimian H, Liu D. A hybrid model for spatiotemporal forecasting of PM2.5 based on wavelet transform and long short-term memory network. Science of The Total Environment, 2019, 664: 1–10.
[11] Li, X., Liu, Y., & Wang, H. (2021). Research on PM2.5 spatiotemporal forecasting model based on LSTM neural network. Environmental Monitoring and Assessment, 193, 1–13.
[12] Bai, Y., Zhuang, Y., & Wang, Y. (2024). Prediction of PM2.5 concentration based on a CNN-LSTM neural network algorithm. Scientific Reports, 14, 1123.
[13] Pedrycz W. Granular computing: An introduction. Human-Centric Information Processing Through Granular Modelling, Springer, 2008: 1–17.
[14] Yao Y, Lin T Y. Granular computing: Basic issues and possible solutions. Rough Sets and Current Trends in Computing, Springer, 1999: 46–57.
[15] Artiemjew, P. (2020). About Granular Rough Computing — Overview of Decision System Approximation Techniques and Future Perspectives. Algorithms, 13(4), 79.
[16] Pedrycz W, Homenda W. Building the fundamentals of granular computing: A principle of justifiable granularity. Applied Soft Computing, 2013, 13(10): 4209–4218.
[17] Yin Y. Prediction and analysis of time series data based on granular computing and support vector machines[J]. Frontiers in Computational Neuroscience, 2023, 17: 1192876.
[18] Song M., Wang R., Li Y. Hybrid time series interval prediction by granular neural network and ARIMA[J]. Granular Computing, 2023, 9(3): 1–12.
[19] Stach W, Kurgan L A, Pedrycz W. Expert-based and computational methods for developing fuzzy cognitive maps. International Journal of Approximate Reasoning, 2005, 39(2–3): 149–192.
[20] Wang, X., et al. (2020). Deep Fuzzy Cognitive Maps for Interpretable Multivariate Time Series Forecasting. IEEE Trans. Fuzzy Systems (2020).
[21] Z. Peng, W. Liu and S. -K. Oh, "AFS-FCM With Memory: A Model for Air Quality Multi-Dimensional Prediction With Interpretability," in IEEE Transactions on Big Data, vol. 11, no. 2, pp. 810-820, April 2025.
[22] Z. Peng, W. Liu, and S. An, “Haze pollution causality mining and prediction with PS-FCM,” Information Sciences, vol. 523, pp. 307–317, 2020.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Academic Journal of Applied Sciences
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
