Deep reinforcement learning-based secondary control for microgrids in islanded mode

Microgrids are generally low-inertia systems with a high penetration of renewable energy sources. The design of advanced control structures is required to keep these grids’ electrical variables within an acceptable range. In this context, the present article proposes an intelligent secondary control...

Descripción completa

Detalles Bibliográficos
Autores: Barbalho, Pedro Inácio de Nascimento e, Albernaz Lacerda Freitas, Vinícius|||0000-0001-8648-9027, Fernandes, Ricardo Augusto Souza, Vinicius Coury, Denis
Tipo de recurso: artículo
Fecha de publicación:2022
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/383524
Acceso en línea:https://hdl.handle.net/2117/383524
https://dx.doi.org/10.1016/j.epsr.2022.108315
Access Level:acceso abierto
Palabra clave:Microgrids (Smart power grids)
Microxarxes (Xarxes elèctriques intel·ligents)
Àrees temàtiques de la UPC::Enginyeria elèctrica
Descripción
Sumario:Microgrids are generally low-inertia systems with a high penetration of renewable energy sources. The design of advanced control structures is required to keep these grids’ electrical variables within an acceptable range. In this context, the present article proposes an intelligent secondary controller for islanded microgrids using the Deep Deterministic Policy Gradient (DDPG). The DDPG controller changes the output power of the storage elements to secure the voltage and frequency stability. This work tested the designed controller for a microgrid that comprises a synchronous generator, two battery energy storage systems and one photovoltaic generator. The controller performance was compared to droop controllers, considering a short-circuit event, feeder and load disconnections. Results showed a consistent reduction of the microgrid’s voltage and frequency deviations with the DDPG algorithm.