Hybrid artificial neural network models for effective prediction and mitigation of urban roadside NO2 pollution

Cabaneros, Sheen Mclean S. and Calautit, John Kaiser S. and Hughes, Ben Richard (2017) Hybrid artificial neural network models for effective prediction and mitigation of urban roadside NO2 pollution. Energy Procedia, 142. pp. 3524-3530. ISSN 1876-6102 (https://doi.org/10.1016/j.egypro.2017.12.240)

[thumbnail of Cabaneros-etal-EP-2017-Hybrid-artificial-neural-network-models-for-effective-prediction-and-mitigation]
Preview
 Text. Filename: Cabaneros_etal_EP_2017_Hybrid_artificial_neural_network_models_for_effective_prediction_and_mitigation.pdf
Final Published Version
License: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 logo
Download (808kB)| Preview

Abstract

Traffic-related air pollution has been a serious concern amongst policy-makers and the public due to its physiological and environmental impacts. An early warning system based on accurate forecasting tools must therefore be implemented to circumvent the adverse effects of exposure to major air pollutants. A multilayer perceptron neural network was trained and developed using air pollution and meteorological data over a two-year period from a monitoring site in Marylebone Road, Central London to predict roadside concentration values of NO2 24 hours ahead. Several hybrid models were also developed by applying feature selection techniques such as stepwise regression, principal component analysis, and Classification and Regression Trees to the neural network model. Most roadside pollutant variables, e.g., oxides of nitrogen, were found to be significant in predicting NO2. The statistical results reveal overall prediction superiority of the hybrid models to the standalone neural network model.

CORE (COnnecting REpositories)