Improving artificial intelligence-based air pollution modeling with the application of meteorological data

A. H. Baghanam, V. Nourani, H. Karimzadeh

    Research output: Contribution to journalArticlepeer-review

    Abstract

    Emotional artificial neural network (EANN) a new generation of artificial neural network linked to the wavelet-based data preprocessing method, was used in this study to predict air pollution in Tabriz city, Iran, from 2015 to 2019. For comparison purposes, the classic feedforward neural network (FFNN) model was applied. Daily meteorological data, as well as pollutants concentration data, were used to predict air pollution concentration in the future. The results showed the efficiency of the EANN model in comparison with classic FFNN in predicting air pollution for Tabriz city. Also, the wavelet EANN (WEANN) model outperformed the wavelet FFNN (WFFNN) and EANN models by up to 4 and 14%, respectively. This result denotes the importance of preprocessing method than applying more updated artificial intelligence methods. The superiority of the proposed WEANN approach is in better learning of extraordinary and extreme values in the validation phase, due to the improved ability of EANN via hormones in comparison with ANN.

    Original languageEnglish
    Pages (from-to)431-446
    Number of pages16
    JournalInternational Journal of Environmental Science and Technology
    Volume21
    Issue number1
    DOIs
    Publication statusPublished - Jan 2024

    Cite this