DEEP REINFORCEMENT LEARNING BASED DYNAMIC DISTRIBUTION NETWORK RECONFIGURATION
DOI:
https://doi.org/10.24867/11BE19PetkovicKeywords:
dynamic network reconfiguration, reinforcement learningAbstract
This paper proposes DDNR based on Reinforcement Learning algorithm. It is multi-objective approach which minimizes cost of energy losses and switching manipulations. The amount of information needed for the algorithm execution is decreased, since the topology information and the information about the power flows in the network are compressed in the single set of variables. This reduces the amount of telemetered measurements needed for the potential real-world execution of the algorithm and it is also convenient from the algorithm training perspective, since the required size of the neural network is reduced.
References
[1] G.Švenda: Specijalizovani softveri u elektroenergetici, skripta sa predavanja iz istoimenog predmeta na master studijama, Fakultet Tehničkih Nauka, Novi Sad, 2019.
[2] Taleski R, Rajicic D. Distribution network reconfiguration for energy loss reduction. IEEE Transactions on Power Systems 1997; 1:398–406.
[3] V.Fathi, H.Seyedi, B.M.Ivatloo: Reconfiguration of distribution systems in the presence of distributed generation considering protective constraints and uncertainties; Int. Trans Electr Energ Systems; February 2020;pp. 1-25; DOI: 10.1002/2050-7038.12346
[4] Kovački NV, Vidović PM, Sarić AT. Scalable algorithm for the dynamic reconfiguration of the distribution network using the Lagrange relaxation approach. International Journal of Electrical Power & Energy Systems 2018; 94:188–202.
[5] Sutton RS, Barto AG. Reinforcement learning: An introduction. MIT press, 2018.
[6] https://searchenterpriseai.techtarget.com/definition/deep-learning-agent 10.09.2020.
[7] Nikolić M, Zečević A. Mašinsko učenje, Matematički fakultet, Univerzitet u Beogradu, 2019.
[8] R. Dugan and T. McDermott. An open source platform for collaborating on smart grid research. Power and Energy Society General Meeting 2011 IEEE 2011; 1–7.
[9] Ramos ER, Exposito AG, Santos JR, Iborra FL. Path-based distribution network modeling: application to reconfiguration for loss reduction. IEEE Transactions on Power Systems 2005; 2:556–564.
[2] Taleski R, Rajicic D. Distribution network reconfiguration for energy loss reduction. IEEE Transactions on Power Systems 1997; 1:398–406.
[3] V.Fathi, H.Seyedi, B.M.Ivatloo: Reconfiguration of distribution systems in the presence of distributed generation considering protective constraints and uncertainties; Int. Trans Electr Energ Systems; February 2020;pp. 1-25; DOI: 10.1002/2050-7038.12346
[4] Kovački NV, Vidović PM, Sarić AT. Scalable algorithm for the dynamic reconfiguration of the distribution network using the Lagrange relaxation approach. International Journal of Electrical Power & Energy Systems 2018; 94:188–202.
[5] Sutton RS, Barto AG. Reinforcement learning: An introduction. MIT press, 2018.
[6] https://searchenterpriseai.techtarget.com/definition/deep-learning-agent 10.09.2020.
[7] Nikolić M, Zečević A. Mašinsko učenje, Matematički fakultet, Univerzitet u Beogradu, 2019.
[8] R. Dugan and T. McDermott. An open source platform for collaborating on smart grid research. Power and Energy Society General Meeting 2011 IEEE 2011; 1–7.
[9] Ramos ER, Exposito AG, Santos JR, Iborra FL. Path-based distribution network modeling: application to reconfiguration for loss reduction. IEEE Transactions on Power Systems 2005; 2:556–564.
Downloads
Published
2020-12-26
Issue
Section
Electrotechnical and Computer Engineering
