Authors: Jonas Van Gompel, Bert Claessens, Chris Develder

Published on: April 23, 2024

Impact Score: 8.0

Arxiv code: Arxiv:2404.14836

Summary

• What is new: The proposal of an ensemble of C-VSNs for system imbalance forecasting, specifically focusing on high imbalance magnitude situations, which are crucial but overlooked in existing methods.
• Why this is important: The increasing challenge and cost of maintaining the balance between electricity generation and consumption due to rising shares of renewables and electrification, and the lack of accurate forecasting models for high imbalance magnitudes.
• What the research proposes: An ensemble of C-VSNs (adaptation of variable selection networks) that predicts the imbalance for the current and upcoming two quarter-hours, including probabilistic forecasts and uncertainty estimations.
• Results: The model outperforms the state-of-the-art by 23.4% for high imbalance magnitude situations and shows a 6.5% improvement in overall forecasting accuracy.

Technical Details

Technological frameworks used: Ensemble of C-VSNs

Models used: Variable Selection Networks (VSNs)

Data used: Imbalance data from Belgium’s TSO, Elia

Potential Impact

This research could disrupt the energy market, particularly benefiting transmission system operators (TSOs) and companies focusing on grid balancing and electricity trading.

Want to implement this idea in a business?

We have generated a startup concept here: GridFlex Analytics.