Artificial intelligence to optimise energy distribution and identify vulnerabilities in distribution networks
Irina Bandura, Mykola Romaniuk, Nataliia Komenda, Andrii Hadai, Liudmyla Konkevych
Streszczenie
The growing complexity and vulnerability of modern power distribution systems require the introduction of inno
vative approaches to their management and protection. The purpose of the study was to substantiate the effectiveness of using
artificial intelligence (AI) as a tool for improving the reliability and manageability of such systems. Within the framework of
the study, modern challenges of electric energy distribution are analysed, software solutions based on AI methods for optimis
ing the distribution of energy resources are implemented, and a comparative analysis of the effectiveness of the corresponding
algorithms in the context of typical threats to power systems is conducted. The main results showed that the use of machine
learning (ML) provided basic load prediction, but did not fully account for peak consumption fluctuations. The clustering
method based on average values revealed characteristic groups of consumers with similar daily profiles, which allowed for
a more accurate prediction of demand and optimised management strategies. On the other hand, the results included an analysis
of the use of AI to identify vulnerabilities in distribution networks. The conclusions emphasised that algorithms for analysing
the technical condition of equipment allow detecting deviations from sensor data without the need for labelled samples. The
results obtained can be used by specialists in the operation of power grids, engineers and software developers to improve the
reliability, stability, and efficiency of managing electrical energy distribution systems in conditions of increasing loads and
various threats.
vative approaches to their management and protection. The purpose of the study was to substantiate the effectiveness of using
artificial intelligence (AI) as a tool for improving the reliability and manageability of such systems. Within the framework of
the study, modern challenges of electric energy distribution are analysed, software solutions based on AI methods for optimis
ing the distribution of energy resources are implemented, and a comparative analysis of the effectiveness of the corresponding
algorithms in the context of typical threats to power systems is conducted. The main results showed that the use of machine
learning (ML) provided basic load prediction, but did not fully account for peak consumption fluctuations. The clustering
method based on average values revealed characteristic groups of consumers with similar daily profiles, which allowed for
a more accurate prediction of demand and optimised management strategies. On the other hand, the results included an analysis
of the use of AI to identify vulnerabilities in distribution networks. The conclusions emphasised that algorithms for analysing
the technical condition of equipment allow detecting deviations from sensor data without the need for labelled samples. The
results obtained can be used by specialists in the operation of power grids, engineers and software developers to improve the
reliability, stability, and efficiency of managing electrical energy distribution systems in conditions of increasing loads and
various threats.