Modeling sulphur dioxide due to vehicular traffic using artificial neural network

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Computational Modeling

Background:

• Vehicular exhaust dispersion is complex and non-linear, challenging traditional deterministic and numerical models.
• Accurate prediction of air pollutants like sulphur dioxide (SO2) is crucial for urban environmental management.
• Existing models struggle to precisely capture the intricate relationships influencing pollutant concentrations.

Purpose of the Study:

• To develop and evaluate an Artificial Neural Network (ANN) model for predicting SO2 concentrations from vehicular sources in urban areas.
• To investigate the impact of combining traffic and meteorological parameters on ANN model performance.
• To demonstrate the capability of ANNs in handling non-linear and noisy data for air quality modeling.

Main Methods:

• An Artificial Neural Network (ANN) model was constructed using various combinations of traffic and meteorological parameters.
• The model was trained and validated to assess its predictive accuracy for SO2 emissions.
• Comparative analysis of model performance based on input parameter sets was conducted.

Main Results:

• ANN models demonstrated a strong ability to recognize and learn from the non-linear patterns in pollutant data.
• The ANN model incorporating both traffic and meteorological parameters exhibited superior performance in predicting SO2 concentrations.
• Individual parameter sets showed varying degrees of predictive power, with combined inputs yielding the best results.

Conclusions:

• Artificial Neural Networks offer a robust approach for modeling complex air pollutant dispersion, outperforming traditional methods.
• Integrating diverse datasets, including traffic and meteorological factors, significantly enhances the accuracy of SO2 concentration predictions.
• The study highlights the potential of ANNs for real-time air quality monitoring and management in urban environments.