DESIGN OF A SYSTEM FOR ANALYSIS AND FORECASTING OF AIR QUALITY

The article addresses the issue of air pollution, which poses serious threats to public health and the environment, according to the World Health Organization. The current state of research and technologies in air quality monitoring and forecasting, as well as existing commercial and scientific projects, are described. It is determined that one of the key tasks is providing accessible information to the public about air pollution levels, enabling timely responses to environmental threats. The article presents the development of an intelligent automated system for monitoring and forecasting air pollution levels. The system is based on modern data analysis methods and neural networks, allowing for pollution level predictions up to six hours in advance. The primary goal of this work is to create a solution that combines the high accuracy of scientific models with accessibility and ease of use for a broad audience. The system provides information in an easy-to-understand format through visualization on a web platform, using Air Quality Indexes (AQI), enabling non-experts to quickly assess health risks. The article also discusses key modern approaches to air pollution forecasting, such as autoregressive models (ARIMAX), regression analysis, and artificial neural networks (ANN), which were employed in the development of the system. The technical aspects of the system are described in detail, including the architecture of the three-tier client-server model, data processing algorithms, and the functional capabilities of the web client for real-time user interaction with the system. The project covers more than 30 cities in Ukraine and collects data from over 100 monitoring stations, allowing for highly accurate air quality forecasting. The web service updates pollution status data every hour, while weather conditions are updated every 15 minutes. By using scalable architecture, the system can handle a large volume of requests and ensure continuous access to up-to-date data. The study's results demonstrated that the developed system can forecast air pollution levels with an accuracy of up to 72% for a six-hour prediction window, confirming the effectiveness of the approach used. The model was tested on real data for Kyiv, and the results showed a high correlation between the predicted and actual air quality values. The proposed system is a promising tool for monitoring and forecasting air quality, contributing to the improvement of environmental conditions and the protection of public health.