A multi-scale evolutionary deep learning model based on CEEMDAN, improved whale optimization algorithm, regularized...

Area of Science:

• Environmental Science
• Data Science
• Machine Learning

Background:

• Air pollution poses significant risks to human health and urban ecosystems.
• Accurate Air Quality Index (AQI) prediction is vital for effective pollution control.
• Traditional AQI forecasting methods exhibit limitations in performance.

Purpose of the Study:

• To propose a novel dual-scale ensemble learning framework for complex AQI time series prediction.
• To enhance the accuracy and reliability of AQI trend forecasting.
• To provide a robust tool for urban air pollution management.

Main Methods:

• Utilized Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Sample Entropy (SE) for AQI series decomposition and reconstruction.
• Employed Long Short-Term Memory (LSTM) neural networks for high-frequency component prediction and Regularized Extreme Learning Machine (RELM) for low-frequency components.
• Optimized LSTM and RELM hyperparameters using an improved Whale Optimization Algorithm (WOA).

Main Results:

• The proposed hybrid prediction model demonstrated significantly improved AQI prediction accuracy across four Chinese cities.
• The dual-scale approach effectively handled the complexity of AQI time series data.
• The CEEMDAN-SE decomposition and WOA-optimized LSTM-RELM ensemble proved effective.

Conclusions:

• The developed dual-scale ensemble learning framework offers a substantial improvement in AQI forecasting accuracy.
• This advanced prediction capability is highly significant for promoting sustainable urban development and public health.
• The methodology provides a reliable approach for addressing complex environmental time series prediction challenges.