Stefan Popović

Sustainability

Time Series Forecasting Methodology for Climatic Drivers of Urban Drought in Sustainable Smart City Planning

Tihi, N., Popov, S., Popović, S., Đukić Popović, S., Samec, N., & Kokalj, F

https://www.mdpi.com/2071-1050/18/8/3945

Urban drought is a climate-related challenge that threatens environmental sustainability, public health, and socio-economic stability in urban areas. With increasing climate variability, sustainable smart city planning requires reliable forecasting methodologies to facilitate adaptive water resource management and long-term climate resilience plans. This study proposes and evaluates a time series forecasting methodology for the climatic drivers of urban drought, using standard statistical approaches—Seasonal Autoregressive Integrated Moving Average ((S)ARIMA) and Holt–Winters exponential smoothing. The methodology includes systematic preprocessing of meteorological data, univariate time series modeling, and performance evaluation using recognized accuracy metrics (RMSE, MAE, and MAPE). Air temperature, precipitation, soil moisture, and wind speed are analyzed as key climatic variables affecting urban drought dynamics. The results indicate that forecast performance varies based on the statistical characteristics of each variable: (S)ARIMA models provide superior predictive accuracy for series with significant seasonality or stochastic fluctuations, whereas the Holt–Winters method is more appropriate for variables displaying sustained downward trends, particularly soil moisture. The forecasts provide a methodological foundation for calculating drought indices and classifying severity, enhancing early warning capabilities and supporting sustainable smart city planning under increasing climate uncertainty.

climate adaptation; climatic drivers; sustainable smart cities; time series modeling; urban drought; water resource management

M22

Veredas Do Direito

Artificial intelligence for environmental risk management in automated boiler systems

Djukic, D., Popovic, S. D., Stojanovic, K., Milić, M., Kihler, M., Milić, S., … Popovic, S.

https://revista.domhelder.edu.br/index.php/veredas/article/view/5823

This paper examines the application of artificial intelligence in automated boiler systems with the aim of improving environmental risk management and reducing emissions generated during combustion processes. The study focuses on the use of neural network models as intelligent monitoring and predictive control tools in industrial heating systems. The main objective of the research is to evaluate how artificial intelligence can support safer and more efficient operation of automated boilers while contributing to lower fuel consumption and reduced environmental impact. The research methodology is based on experimental data collected from an automated boiler system of the OZON 55 type equipped with sensor-based monitoring devices. Operational parameters such as temperature, air supply, fuel characteristics, and gas emissions were recorded and analyzed using recurrent neural network models designed to predict deviations in combustion behavior. The obtained results indicate that neural network–based predictive monitoring can detect anomalies in operational parameters at an early stage and enable timely adjustments of combustion conditions. Such improvements contribute to increased operational safety, improved fuel efficiency, and lower emissions of harmful gases. The findings suggest that the integration of artificial intelligence into automated boiler systems represents an effective technological approach for enhancing environmental protection, improving risk management, and supporting more sustainable energy use in industrial heating systems.

Neural Networks, Risk Management, Automated Boiler Systems, Environmental Protection

M23

Studies in Informatics and Control

A Hybrid MCDM Framework for Selecting Optimal AI Algorithms in Real-Time Infrared Signal Detection Systems

Nikola GLIGORIJEVIĆ, Dejan VIDUKA, Stefan POPOVIĆ, Danilo STRUGAREVIĆ, Vladimir ČABRIĆ

https://sic.ici.ro/current-issue/a-hybrid-mcdm-framework-for-selecting-optimal-ai-algorithms-in-real-time-infrared-signal-detection-systems/

This paper proposes a hybrid multi-criteria decision-making (MCDM) framework for selecting the optimal AI algorithms in the context of real-time infrared signal detection systems. Five performance criteria were considered, namely the processing speed, detection accuracy, segmentation efficiency, noise robustness and energy efficiency, reflecting the requirements of real-time image processing and embedded computer vision systems. This framework integrates the SWARA method for expert-based criteria weighting with Net Worth Analysis (NWA) for algorithm ranking, enabling a transparent and systematic evaluation. The experimental results show that the Fast R-CNN algorithm achieves the highest overall performance, while algorithms such as EfficientDet obtain lower scores and require further refinement to be effectively used in real-time infrared signal detection applications. To sum up, the proposed method addresses the current lack of structured decision-support tools for selecting among various AI-based infrared signal detection models under operational constraints. The research findings provide actionable guidance for researchers and practitioners developing embedded AI, surveillance and automated monitoring systems.

Multi-criteria decision making (MCDM), SWARA method, Net Worth Analysis (NWA), Artificial intelligence, Image processing algorithms, Computer vision, Algorithm evaluation

M23

REVIZOR * Časopis Za Upravljanje Organizacijama, Finansije I Reviziju

Emocionalna inteligencija i percepcije ekonomskih performansi u organizacijama u Srbiji.

Miličić, B., Vukonjanski Srdić, J., & Popović, S.

https://casopisrevizor.rs/index.php/revizor/article/view/241

Cilj istraživanja je ispitivanje odnosa između emocionalne inteligencije zaposlenih i njihovih percepcija ekonomskih performansi organizacije. Rezultati ukazuju na statistički značajnu, ali umerenu povezanost emocionalne inteligencije i ekonomskih performansi, uz izražene razlike u obrascima povezanosti između poduzoraka formiranih prema nivou emocionalnih kompetencija. Nalazi potvrđuju da emocionalna inteligencija u organizacijama u Srbiji ima ograničen, ali relevantan doprinos u objašnjenju ekonomskih procena zaposlenih, pri čemu njen značaj zavisi od individualnih i organizacionih uslova.

emocionalna inteligencija, ekonomske performanse, percepcije zaposlenih, organizacioni ishodi, organizacije u Srbiji

M52

Electronics

Optimization of Artificial Intelligence Algorithm Selection: PIPRECIA-S Model and Multi-Criteria Analysis

Popović, S., Viduka, D., Bašić, A., Dimić, V., Djukic, D., Nikolić, V., & Stokić, A.

https://www.mdpi.com/2079-9292/14/3/562

In the age of digitization and the ever-present use of artificial intelligence (AI), it is essential to develop methodologies that enable the systematic evaluation and ranking of different AI algorithms. This paper investigated the application of the PIPRECIA-S model as a methodological framework for the multi-criteria ranking of AI algorithms. Analyzing relevant criteria such as efficiency, flexibility, ease of implementation, stability and scalability, the paper provided a comprehensive overview of existing algorithms and identified their strengths and weaknesses. The research results showed that the PIPRECIA-S model enabled a structured and objective assessment, which facilitated decision-making in selecting the most suitable algorithms for specific applications. This approach not only advances the understanding of AI algorithms but also contributes to the development of strategies for their implementation in various industries.

PIPRECIA-S model; artificial intelligence; multi-criteria analysis; algorithm ranking; strategic planning

M22

International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE)

A Hybrid SWARA-NWA Framework for Evaluating AI-Based Image Recognition Algorithms in Educational Technology Applications.

Gligorijević N., Djukić Popović S., Nikolić V., Viduka D. & Popović S.

https://ijcrsee.com/index.php/ijcrsee/article/view/3377

Artificial Intelligence (AI) and computer vision technologies are increasingly integrated into educational environments through intelligent tutoring systems, gesture-based learning, facial expression analysis, and automated evaluation tools. However, selecting the most appropriate image recognition algorithms for educational applications remains a challenge due to varying requirements regarding speed, accuracy, hardware compatibility, and usability in dynamic classroom conditions.This paper proposes a hybrid multi-criteria decision-making (MCDM) model based on the Step-wise Weight Assessment Ratio Analysis (SWARA) and Net Worth Analysis (NWA) methods to evaluate and rank nine widely used AI-based visual recognition algorithms. The evaluation is conducted using five education-relevant criteria: processing speed, recognition accuracy, robustness to classroom noise, compatibility with low-end devices, and energy efficiency. Expert assessments from the field of educational technology were used to derive weight coefficients and evaluate algorithm performance.The results show that Fast R-CNN achieved the highest overall score (1.141), followed by U-Net (1.077) and DeepLab (1.062), indicating their suitability for real-time and resource-constrained EdTech environments. Algorithms such as MobileNet (1.057) and YOLO (1.037) also demonstrated balanced performance, making them viable for mobile or moderately demanding educational scenarios. The proposed model offers a structured and transparent decision-support framework that can assist researchers and practitioners in selecting optimal AI algorithms for diverse educational applications.

Educational technology, Artificial intelligence, Computer vision, Multi-criteria decision making (MCDM) and Algorithm evaluation

M22

Journal of Soft Computing and Decision Analytics

Risk Management Innovations through Neural Network Integration in Automated Boiler Combustion Systems.

Popovic, S., Popovic, S. D., Denic, N., Djukic, D., & Stojanovic, J.

https://jscda-journal.org/index.php/jscda/article/view/65

In the early decades of the twenty-first century, the application of artificial intelligence has been expanding across all sectors of society, including industrial energy systems. This paper emphasizes the significance of integrating artificial neural networks into boilers with automatic firing, as part of a research project currently in its fifth year of experimental validation. The implementation of neural networks in such systems has demonstrated promising results in the domain of risk management, particularly through the prediction of system malfunctions and their proactive elimination via software interventions. The application of AI-based solutions in boiler control not only contributes to the reduction of environmental impact but also enhances operational safety by preventing accidents that may endanger human health and cause material losses.

Neural networks , Risk management, Automatic boilers, Combustion optimization

SJR Q1

Proceedings of IRASA International Scientific Conference Science, Education, Technology and Innovation SETI V 2023, Belgrade, Serbia

Student discipline as today’s social security problem – the role of the education system in removing this problem

Stefan Popović, Jovan Ničković, Sonja Đukić Popović, Vladimir Čabrić, Jovan Veselinović, Milan Gligorijević, Dejan Đukić

—

Student indiscipline is one of the major problems not only of education, but also of modern society as a whole. The educational system represents the first contact of the youth with the state apparatus, and then, at the earliest age, it is the link that turns a child into a man with all social responsibilities and obligations. The work deals with identifying the key causes of student indiscipline and finding ways to prevent them. Students, teachers and parents of several primary and secondary schools of the school administrations of Belgrade and Niš played a major role in the preparation of the paper.

student indiscipline, national security, upbringing, education

M33

Proceedings of 9th Virtual International Conference on Science, Technology and Management in Energy, Belgrade, Serbia

Neural networks in pellet combustion control - an overview of the group's research work in 2022/2023

Stefan Popovic, Dejan Djukic, Sonja Djukic Popovic, Milan Gligorijevic,

https://energetics.cosrec.org/previous-energetics/

The problems of pollution and global warming have plagued the planet for more than a century, and are the result of excessive consumption of fossil fuels. The last decades have brought innovations in the heating of smaller buildings, heating fuel oil and coal are being shut down, and boilers with automatic gas and biomass heating are being introduced. This significantly reduces pollution, but not enough. Hence the need for greater application of artificial intelligence and machine learning in combustion control in boilers with automatic firing. This paper presents a description of the experimental application of artificial intelligence, machine learning and neural networks to the ATI Terming Ozone 55 boiler and a brief summary of the results obtained.

boilers with automatic firing, neural networks, environmental protection

M33