부산대학교 도서관

The article deals with the problem of air pollution near objects whose direct activity is the cause of atmospheric pollution. The presence of a pollutant in the atmospheric boundary layer has a negative impact both on flora and fauna. The authors set the goal of reducing the probability of a man-made threat to such objects. To do this, the historical data of the real object of the loading zone of the transport and logistics hub were analyzed. Developed a method of pollutant concentration prediction using machine learning methods. The method allows one to predict the concentration near sensors at discrete time intervals from 1 hour to 24 hours. Also, a problem that allows one not only to predict the concentration near the sensors but also over the entire controlled area of the object was solved by using the Gaussian plume model and gradient boosting of decision trees.