Prediction of Air Pollution by Particulate Dust Particles of PM2.5 and PM10 Using an Artificial Intelligence-Based Method

The article analyzes the development of specialized software for predicting PM2.5 and PM10 air pollution levels using an artificial intelligence approach. The core objective is to develop an algorithm and its corresponding software based on neural networks. This software aims to quickly and accurately analyze air quality with high accuracy and speed within specific urban or industrial zones, focusing on particulate matter with diameters of 2.5 and 10 µm. This predictive capability will ultimately affect the decision-making processes to mitigate or minimize air pollution levels. The study focuses on the development of the prediction algorithm itself (object) and the relevant software utilizing neural networks to achieve this goal (subject). The chosen XGBoost models (XGBoost-linear-regression and XGBoost-mlp) demonstrate exceptional accuracy (average absolute error below 1.47%) and efficiency, paving the way for informed actions to lessen the environmental and public health impacts of air pollution.