Related Experiment Videos
Two-stage deep learning hybrid framework based on multi-factor multi-scale and intelligent optimization for air
Jujie Wang1,2, Wenjie Xu1, Jian Dong1
1School of Management Science and Engineering, Nanjing University of Information Science and Technology, Nanjing, 210044 China.
Stochastic Environmental Research and Risk Assessment : Research Journal
|April 4, 2022
Summary
This study presents a novel framework for predicting particulate matter (PM2.5) concentrations and providing early warnings. The advanced hybrid model demonstrates superior accuracy and stability in air pollution forecasting.
Area of Science:
- Environmental Science
- Data Science
- Public Health
Background:
- Accurate prediction of air pollution, particularly PM2.5, is crucial for public health and urban environmental management.
- Existing models often struggle with the complexity and variability of air pollutant data.
Purpose of the Study:
- To develop and validate an innovative hybrid framework for enhanced air pollutant prediction and early warning systems.
- To improve the accuracy and reliability of PM2.5 forecasting.
Main Methods:
- Utilized Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and fuzzy entropy for signal decomposition and reconstruction.
- Employed the Max-Relevance and Min-Redundancy (mRMR) method for feature selection.
- Developed a two-stage deep learning model combining Long Short-Term Memory (LSTM) with Gray Wolf Optimization (GWO) for prediction and nonlinear integration.
Main Results:
- The proposed hybrid framework achieved superior prediction accuracy, warning accuracy, and prediction stability compared to other models in empirical studies across three Chinese cities.
- The model effectively captures complex temporal dependencies and nonlinear relationships in PM2.5 data.
Conclusions:
- The developed hybrid framework serves as an effective tool for real-time air pollutant prediction and early warning.
- This approach offers a significant advancement in air quality monitoring and management strategies.
