RELY Nominated for the Staatspreis Technologie 2024
What does Reinforcement Learning have to do with power grids? Read more about RELY and find out!
![Clemens Heitzinger](/news/150/clemens-heitzinger_hu5e011b906fbd25593632515f1a3a4e0a_294010_770x550_fill_q80_bgffffff_lanczos_smart1_2.jpg)
We are delighted to announce that the project RELY was nominated for this year’s Staatspreis Technologie!
The research project RELY, lead by Clemens Heitzinger, Co-head of CAIML, and Felix Birkelbach, was nominated for the Staatspreis Technologie 2024 (National Award for Technology) in the category “AI for Green”. On June 11, the project was honored at a gala hosted by the Bundesministerium für Klimaschutz, Umwelt, Energie, Mobilität, Innovation & Technologie (Ministry for Climate Action, Environment, Energy, Mobility, Innovation and Technology).
RELY is a collaboration of experts in automation technology and mechanical engineering that applies RL to energy systems. RELY develops Reinforcement Learning (RL) algorithms to control highly complex processes in energy systems. The innovative use of digital twins to train these algorithms increases their reliability. The methods developed are tested on a reversible pump turbine in a lab and are scalable to various power-generating units. The algorithms that RELY develops have immense potential for flexibility, an essential quality for energy systems of the future.
To keep our energy grid stable, even with a high percentage of renewable energy sources, our systems need to respond quickly to fluctuations. This is where reinforcement learning (RL) comes in. RL learns the best control strategies directly from system data and interactions, which makes it possible to manage energy systems more accurately and dynamically than conventional methods do. Reinforcement algorithms are incredibly adaptive, able to process vast amounts of information and learn continuously.
One of the biggest hurdles is ensuring the reliability of these algorithms. To address this issue, a lab-scale system at TU Wien, featuring a 400-kilowatt pump turbine, is used. The setup mimics a large-scale plant, which allows for the effective training and testing of the algorithms. Pump turbines are crucial, as they generate electricity or store energy when there’s an oversupply, and they also play a role in regulating the grid. While RELY’s current focus is on hydropower, energy systems of the future will be highly diverse and include technologies like photovoltaics and wind power. RL gives us a chance to shape the landscape of future energy systems, making them more efficient and resilient.
What does Reinforcement Learning have to do with power grids? Read more about RELY and find out!
If you watch this video, data may be transmitted to third parties.
Watch on YouTube: https://youtu.be/VErNKEz98zg