Application of Artificial Intelligence Algorithms in Manufacturing Processes and Production Planning

The project was developed with the goal of improving the reliability and efficiency of critical equipment at the Nikola Tesla thermal power plant, in collaboration with Elektroprivreda Srbije (EPS) and the Mihajlo Pupin Institute. The focus of the project is the development of an AI-based predictive maintenance model that enables early detection of potential failures and reduction of downtimes, thereby optimizing plant operation and extending equipment lifespan.

The model is based on advanced anomaly detection algorithms that analyze multidimensional time series of historical data collected from industrial sensors. During training, the model learns patterns of normal equipment behavior, and uses statistical methods and machine learning techniques to detect deviations from standard operational parameters. Once trained, the deployed model enables real-time anomaly detection by processing continuously collected data from SCADA system. The model has been applied to monitor critical components of thermal power plant infrastructure, including boiler feedwater pump, mills, forced draft and flue gas fans at the Nikola Tesla power plant, as well as the conveyor gearbox in the Kolubara mining basin.

The developed AI model is intended for engineers and operators at the thermal power plant, as well as maintenance teams within EPS. Its application enables more accurate planning of maintenance interventions, reduction of unplanned outages, and improved operational efficiency. In the long term, the system contributes to lower maintenance costs and enhanced energy security.