Vicomtech obtains first place in the challenge of Demand Forecasting in the Low Voltage network within the framework of the Innovation Data Space i-DS project in collaboration with Iberdrola

Vicomtech has been the winner of the i-DS use case, Demand Forecasting in the Low Voltage (LV) network for the generation of future consumption scenarios.

Vicomtech collaborates with Iberdrola in the Innovation Data Space (i-DS) project with the general objective of streamlining and enhancing the extraction of value from data to improve the service of distribution networks through Artificial Intelligence. This virtual data laboratory is part of the Global Smart Grids Innovation Hub - Iberdrola, a global reference innovation center in smart grids of Iberdrola Distribución España, in which Vicomtech has been involved since 2021.

For the resolution of the challenge, Vicomtech has developed demand prediction models and short-term (hourly, up to 7 days) and medium-term (daily, up to 4 months) electricity generation models at the Low Voltage Line (LBT) and Transformer Center (CT) levels, based on data from 77,000 CUPS. Vicomtech has carried out an exhaustive analysis of the quality of the data, detecting the presence of gaps in the consumption curves, especially at supply points with self-consumption contracts. Advanced predictive models were implemented, using weather and work data to improve the accuracy of predictions. The methodology followed included gap data imputation, CUPS data aggregation, anomaly detection, and clustering to train efficient predictive models.

The results showed that the short-term predictive models of imported energy are enough accurate, while the long-term models are also notable, although they would benefit from a longer history. The report carried out within the framework of this collaboration also addresses the impact of self-consumption, electric vehicles and other policy changes on demand predictions. Improvements in the reliability of the models and the automation of predictions through change point detection algorithms are also proposed. Finally, possible future work is highlighted, such as obtaining missing and more accurate data to improve Activa Export predictions and implementing prediction automation algorithms.

The collaboration between Vicomtech and Iberdrola goes further. Another initiative in which they participate jointly is the IA4TES project - Artificial Intelligence for the Sustainable Energy Transition, led by Iberdrola, which is part of the R&D Missions Program in Artificial Intelligence, under the Digital Spain 2025 Agenda and the Strategy National Artificial Intelligence.