2025/07/02 by Manzhilevskaya, S. E., Mailyan, D. R.
#air pollution #artificial intelligence #dust pollution #environmental safety of urban areas #fine dust #загрязнение воздушной среды #искусственный интеллект #мелкодисперсная пыль #пылевое загрязнение #экологическая безопасность городских территорий
paper · doi:10.71536/sd.2025.2c31.12
Currently, the construction industry is actively implementing advanced digital technologies, which opens up new opportunities for studying the potential of predictive tools in the field of air pollution control with fine dust in order to improve the environmental safety of urban areas. Using existing artificial intelligence algorithms, it is possible to effectively track the concentration of dust particles in the Earth's atmosphere. To confirm the possibility of long-term forecasting of dust pollution during construction work, seven machine learning models were tested: ARIMA, EMA, Prophet, NARX and NNAR neural networks, Random Forest, SVM and XGBoost. The aim of the research was to evaluate the effectiveness of predicting the level of dust pollution using various machine learning models. Using the Modeltime software, a detailed analysis of correlations between meteorological parameters and concentrations of fine particles was carried out. The results of the study show that the use of ensemble modeling provides effective forecasts of the level of atmospheric air pollution. Among the seven tested machine learning algorithms, the most accurate in predicting the concentration of fine particles were identified - ARIMA, Random Forest and XGBoost.