An AQI decomposition ensemble model based on SSA-LSTM using improved AMSSA-VMD decomposition reconstruction technique

Area of Science:

• Environmental Science
• Data Science
• Artificial Intelligence

Background:

• Air quality index (AQI) is crucial for monitoring air pollution and public health.
• Accurate AQI prediction is essential for timely air pollution control and management.

Purpose of the Study:

• To develop a novel integrated learning model for enhanced AQI prediction.
• To improve the accuracy and stability of AQI forecasting using advanced optimization and decomposition techniques.

Main Methods:

• An improved Artificial Marine-Songbird Algorithm (IAMSSA) was developed for optimizing Variational Mode Decomposition (VMD) parameters.
• IAMSSA-VMD decomposed non-linear AQI data into regular sub-sequences.
• A Sparrow Search Algorithm (SSA) optimized Long Short-Term Memory (LSTM) parameters for each decomposed component, creating an ensemble model (IAMSSA-VMD-SSA-LSTM).

Main Results:

• IAMSSA demonstrated superior convergence, accuracy, and stability compared to conventional algorithms.
• The IAMSSA-VMD-SSA-LSTM model achieved optimal prediction performance with MAE, RMSE, MAPE, and R² values of 3.692, 4.909, 6.241, and 0.981, respectively.
• The proposed model exhibited superior generalization ability compared to other tested models.

Conclusions:

• The IAMSSA-VMD-SSA-LSTM decomposition ensemble model offers higher prediction accuracy and improved fitting and generalization capabilities.
• This study provides a robust theoretical and technical foundation for air pollution prediction and ecosystem restoration efforts.