Loading...
Loading...
Contributors:
Dataset URI:
10.5281/zenodo.22706976Synthetic time series of aggregated national electricity consumption (Portugal), at 15-minute resolution, generated by a diffusion model conditioned on weather and calendar context. The dataset contains 608,160 records, corresponding to 1,267 days with five independent realizations each, covering the period from January 1, 2023 to June 21, 2026. Each day is generated five times from the same context: since diffusion sampling starts from random noise, the five realizations are distinct and equally plausible samples from the distribution of days compatible with that context, which allows the dataset to be used in studies requiring variability rather than a single scenario. The data is synthetic. It contains no personal information and does not constitute observations of the electrical system. It is intended for studying and simulating consumption scenarios, evaluating flexibility management strategies, and testing market mechanisms, in contexts where real data is not accessible for reasons of availability or privacy. Regarding quality: a classifier trained to distinguish real from synthetic series performs essentially at chance level, with a discriminative score of 0.016 ± 0.012 across ten independent runs. The DATASHEET.md file documents the provenance, column structure, full evaluation, and known limitations. The generative model is a Diffusion-TS augmented with a retrieval-augmented generation (RAG) mechanism, trained on the Deucalion supercomputer (EuroHPC/MACC) using real consumption data from E-REDES and temperature data from Open-Meteo, both under CC BY 4.0 license. Produced within the scope of the master's dissertation "Contextual generative models for time series using high-performance computing," Universidade de Trás-os-Montes e Alto Douro, 2026.
Keyword(s):
computação de alto desempenho; consumo elétrico; dados sintéticos; diffusion models; electricity consumption.
MeSH Term(s):
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
No Associated Publication Data Available
Supplementary Data:
No Supplementary Data Available
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)