A Sequence-to-sequence LSTM aproach for forecasting energy consumption and production

Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26220%2F26%3A0198046" target="_blank" >RIV/00216305:26220/26:0198046 - isvavai.cz</a>
Result on the web
<a href="http://dx.doi.org/10.1109/EPE67256.2025.11127157" target="_blank" >http://dx.doi.org/10.1109/EPE67256.2025.11127157</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.1109/EPE67256.2025.11127157" target="_blank" >10.1109/EPE67256.2025.11127157</a>

Result language
angličtina
Original language name
A Sequence-to-sequence LSTM aproach for forecasting energy consumption and production
Original language description
As the energy sector is undergoing a major transformation due to innovation and the growing share of new renewable energy sources, accurate forecasting of electricity production and consumption is crucial for efficient market management and optimization. This paper focuses on the application of advanced deep learning methods to the behaviours of prosumers. The importance of prosumers lies in knowing their own consumption and production in order to effectively utilize their role in the energy sector. To enhance the understanding of consumption and production patterns, a Long Short-Term Memory network with a sequence-to-sequence architecture was implemented to forecast the consumption and production of prosumers. The model was trained and evaluated on real data from residential photovoltaic systems and smart meters over a four-year period. The results show that the model can accurately predict values based on historical data. These models demonstrate the potential of deep learning for improving local energy management and enabling more active participation of prosumers in electricity markets. Future areas of research should focus on how prosumers can interact with the market and what options they have for entering the market to achieve some profit.
Czech name
—
Czech description
—

Publication year
2025
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection
Proccedings of the 2025 25th International Scientific Conference on Electric Power Engineering (EPE)
ISBN
979-8-3315-8636-2
ISSN
—
e-ISSN
—
Number of pages
4
Pages from-to
1-4
Publisher name
IEEE
Place of publication
New York
Event location
Praha
Event date
May 27, 2025
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
001575469500035

Similar results(10)