Smart grid optimized operation driven by reinforcement learning
This thesis focuses on the development of a reinforcement learning model for the operation and demand response control of a smart grid. First, a generic problem is formulated to define the demand response control. Then a study case is proposed with different locations of distributed energy resources...
| Autor: | |
|---|---|
| Formato: | tesis de maestría |
| Fecha de publicación: | 2022 |
| País: | España |
| Recursos: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/374273 |
| Acesso em linha: | https://hdl.handle.net/2117/374273 |
| Access Level: | acceso abierto |
| Palavra-chave: | Smart power grids Xarxes elèctriques intel·ligents Àrees temàtiques de la UPC::Matemàtiques i estadística |
| Resumo: | This thesis focuses on the development of a reinforcement learning model for the operation and demand response control of a smart grid. First, a generic problem is formulated to define the demand response control. Then a study case is proposed with different locations of distributed energy resources and flexible components for reducing the cost associated with its grid con- sumption and safety management. The potential application of different deep reinforcement learning models with different activation functions and network shapes, among them, will be compared and analysed for the grid operation. The goal is to find a deep reinforcement learning model to optimize the demand side of energy management of a smart grid, that achieves better results than other existing approaches. Finally, a new policy for deep reinforcement learning algorithms will be proposed. This will provide a tool to guide the energy management of elec- trical distribution grids with high penetration of renewable energy sources. |
|---|