Застосування штучного інтелекту в наукових дослідженнях у галузі теплоенергетики: можливості, виклики та етичні аспекти

Abstract

The article provides a comprehensive analysis of the current state and prospects for the application of artificial intelligence (AI) technologies in scientific research and educational purposes in the field of thermal power engineering, taking into account technical, methodological, and ethical and legal aspects. Four key areas of use of machine learning methods and neural networks are considered: forecasting and optimization of thermal processes, predictive maintenance of thermal power equipment, optimization of operating modes of thermal generating units, as well as computer modeling of combustion and control of emissions of harmful substances. It is shown that physically informed neural networks (PINN) and hybrid approaches ensure the physical correctness of results even with a limited amount of training data. It is established that predictive maintenance systems are able to reduce unplanned equipment downtime and reduce repair costs. Special attention is paid to methodological challenges: problems of data quality and representativeness, interpretability of models, and correct validation of time series. The requirements of the EU Regulation on Artificial Intelligence (EU AI Act 2024/1689) for AI systems in critical infrastructure are analyzed – in particular, the requirements for risk management, algorithm transparency and human oversight. The prospects of the concept of a digital twin of a thermal energy facility and the role of generative AI in accelerating scientific research are substantiated. Systemic barriers to the implementation of AI solutions in the industry are identified and the need for interdisciplinary training of specialists is emphasized. It was determined that promising areas of further research are the development of standardized methodological protocols for validating AI models in the thermal power industry, the development of methods for applying AI for applied problems in the industry, as well as the study of organizational and ethical aspects of implementing AI solutions at energy enterprises.