STREAM partners publish new paper on improving energy autonomy of Positive Energy Districts with AI

A new paper titled “Improving energy autonomy of positive energy districts using multi-agent deep reinforcement learning” has been published in Scientific Reports (Nature Portfolio). The work was co-authored by STREAM partner Comsensus together with researchers from the Jozef Stefan Institute (IJS).

The study examines how advanced Artificial Intelligence (AI) methods specifically multi-agent deep reinforcement learning (MADRL) can improve the operation of Positive Energy Districts (PEDs). PEDs are designed to generate more renewable energy than they consume and are considered an important step towards climate-neutral cities in Europe.

The results show that MADRL can coordinate distributed energy resources such as photovoltaic panels, batteries, and electric vehicles more effectively. This enables PEDs to increase energy self-sufficiency, reduce reliance on external grids, and improve resilience to fluctuations in renewable generation.

The full article is available here.