Commercial and public buildings have complex energy use patterns shaped by weather, occupancy, schedules and equipment. Traditional benchmarking produces shallow comparison because it relies on limited normalization (e.g. simple weather corrections) or expensive submetering systems that are difficult to scale. Machine learning offers a robust alternative that captures the combined effects of weather, time and operational patterns to estimate system-level loads and create valuable and actionable performance baselines.

AimBees turns the high-resolution data of smart meters into operational energy-efficiency benchmarks. Instead of comparing annual or monthly consumption, the project separates a building’s load into interpretable load signatures – such as baseload, shutdown depth, weekend reduction or start-up time – and derives building-specific ideal energy profiles from comparable buildings in the same portfolio.

The result is a service that estimates realistic operational saving potential and delivers highly automated building reports together with expert-supported portfolio recommendations. It supports building owners, portfolio managers, consultants and public authorities, and contributes to the implementation of the Energy Performance of Buildings Directive (EPBD).

The approach will be validated on historical datasets and real pilot buildings, and the solution will be advanced from TRL 7 to a market-ready service (TRL 9).

• Eurostars project 11009, running from September 2026 to December 2028
• Consortium of four partners from two countries
  • Czech Republic: SEVEn, Energy Twin
  • Austria: TU Wien, e-think energy research
• Validation on 400 buildings, three detailed case studies and thirty short feedbacks from users

Coordinator

Michal Staša
SEVEn, The Energy Efficiency Center
michal.stasa@svn.cz
+420 724 992 441