This paper presents the results of a study aimed at the development of mathematical methods for assessing air pollution based on fuzzy logic and neural network technologies. The object of the study is the dispersion of vehicle-related harmful substances, and the subject is the patterns of distributing the concentrations of these substances under the influence of urban area factors and weather conditions. The goal of the study is to develop a mathematical model for the dynamic calculation of the pollutant concentration cloud using computer modeling. The developed methods, as opposed to the existing ones, allow assessing the distribution of emission concentrations in real time and take into account the influence of building geometry factors, wind shadows, and weather conditions. The proposed approach allows detailing the spatial heterogeneity of air pollution in densely populated areas. The modeling results showed that under certain development parameters, the emission concentration in the leeward zone of buildings can more than double as compared to open urban environment areas. The analysis of the obtained data showed that the deviation of the results as compared to laboratory measurements does not exceed 20% in most of the studied urban areas, which confirms the high accuracy of the model. The results of the study have found their practical application as an algorithm integrated into the AIMS eco software which can be used for real-time environmental monitoring and the development of measures to reduce urban air pollution.
Shepelev et al. (Wed,) studied this question.