Artificial Intelligence in Predicting Emissions is the central topic of this scientific publication.
Abstract
The Bor mining complex represents one of the most significant industrial installations in Serbia, characterized by a long history of intensive metallurgical and mining activities and a substantial impact on local air quality and environmental safety. As an IPPC installation, the complex is subject to strict regulatory requirements related to integrated pollution prevention, continuous emission monitoring, and risk management. However, traditional monitoring systems for sulfur dioxide (SO₂), particulate matter (PM10 and PM2.5), and heavy metal emissions—primarily based on reactive measurement and post-incident reporting—have demonstrated limited effectiveness in preventing environmental incidents and recurring exceedances of regulatory standards. Frequent episodes of elevated pollution levels indicate the need for more advanced, predictive, and preventive approaches to environmental management. This paper examines the role of digital technologies and artificial intelligence (AI) in enhancing emission prediction and early detection of environmental incidents within IPPC installations, with the Bor mining complex serving as a representative case study. The research is based on the analysis of publicly available datasets obtained from governmental environmental monitoring systems, university-based measurement series, and real-time air quality index (AQI) monitoring platforms. These datasets provide long-term and high-frequency time-series data suitable for advanced analytical processing. The study explores the application of AI-based methods, including time-series analysis and predictive algorithms such as Long Short-Term Memory (LSTM) neural networks, for forecasting emission trends and identifying conditions that may lead to environmental incidents. The results indicate that AI-supported models can significantly improve the capacity for early warning, risk assessment, and decision support, thereby enabling a transition from reactive compliance toward proactive environmental protection. In addition to emission control, the proposed approach contributes to improved environmental safety, protection of public health, and increased transparency of IPPC compliance. Based on the findings, the paper proposes an integrated smart environmental monitoring framework that combines continuous digital monitoring with AI-driven predictive analytics. This framework may serve as a reference model for other high-risk industrial installations in Serbia and the wider region, supporting more effective implementation of IPPC requirements and strengthening the link between environmental regulation, technological innovation, and disaster risk reduction.
How to cite
Gajović, A., Cvetković, V. M., & Milenković, D. (2026). Veštačka inteligencija u predikciji emisija i ekoloških incidenata u IPPC postrojenjima: Rudnik Bor kao studija slučaja [Artificial intelligence in predicting emissions and environmental incidents in IPPC installations: The Bor mine as a case study]. U I. Ilić, D. Barjaktarević, & C. Smilevski (Ur.), Stanje i zaštita životne sredine – multidisciplinarni pristup: Zbornik radova (str. 279–296). Beograd: Fakultet za informacione tehnologije i inženjerstvo, Univerzitet „Union – Nikola Tesla“.
