Arithmetic optimization algorithm with deep learning enabled airborne particle-bound metals size prediction model.

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Data Science

Background:

• Heavy metal air pollution from airborne particulate matter (APM) poses significant health risks.
• Understanding the size distribution of airborne heavy metals is crucial for accurate health risk assessment.
• Deep learning models offer advanced prediction capabilities compared to traditional methods.

Purpose of the Study:

• To develop and evaluate a novel deep learning model for predicting size-fractionated airborne particle-bound metals.
• To accurately forecast the concentration of particulate matter (PM) and distinct sized-fractionated APM.
• To determine the temporal trends of atmospheric heavy metals.

Main Methods:

• A novel Arithmetic Optimization Algorithm (AOA) integrated with a Multi-Head Attention based Bidirectional Long Short-Term Memory (MABLSTM) model was proposed.
• The AOA was employed for optimal hyperparameter tuning of the MABLSTM model.
• The model's predictive accuracy was validated through comparative studies with existing methods.

Main Results:

• The AOA-MABLSTM model demonstrated superior performance in predicting size-fractionated airborne heavy metals.
• Specific Root Mean Square Error (RMSE) values were achieved for Aluminum (73.200), Copper (6.747), and Zinc (lowered by 45.250).
• Simulation results confirmed the model's effectiveness over other contemporary approaches.

Conclusions:

• The proposed AOA-MABLSTM model provides an accurate and effective approach for forecasting airborne heavy metal concentrations.
• This advancement aids in better understanding and mitigating the health impacts of heavy metal air pollution.
• The study highlights the potential of advanced deep learning techniques in environmental monitoring and analysis.