Dumont, Axel (2024) Phase balancing policy in low-voltage distribution network by reinforcement learning PFE - Project Graduation, ENSTA.

[img]
Preview
PDF
3510Kb

Abstract

Load imbalance in low-voltage electricity distribution networks is a critical problem that leads to inefficiencies and high technical losses. This problem arises mainly from the uneven distribution of single-phase consumers in the three-phase system, and from inhomogeneous loads on each phase. Resolving this imbalance offers numerous benefits, including reduced power losses, lower neutral wire current, prevention of overheating and premature ageing of transformers, avoidance of equipment malfunctions and better utilization of existing electrical capacity. These improvements not only enhance the overall efficiency of the distribution network, but also contribute to improved service quality and cost savings for utilities. This would also benefit and increase the capacity of distributed energy resources (DER) such as solar photovoltaic. This work allowed us to first observe the state of the art of this problem by a bibliographical search and to compare results with other AI methods. In a second stage, this internship allowed to model the problem of load balance by developping an environment of reinforcement learning. Finally, tests with pre-implemented models were carried out in order to observe their performance in the previous environment, opening the door for future research that will easily take the problem into hand thanks to this information.

Item Type:Thesis (PFE - Project Graduation)
Uncontrolled Keywords:Reinforcement Learning, Low-voltage distribution network, Artificial Intelligence, Graph Neural Network, Optimization.
Subjects:Information and Communication Sciences and Technologies
ID Code:10383
Deposited By:Axel DUMONT
Deposited On:24 sept. 2024 13:04
Dernière modification:24 sept. 2024 13:04

Repository Staff Only: item control page