What the study found
The study found that conditional diffusion models can synthesize plausible daily active and reactive power profiles for low-voltage distribution substations. The authors report that the generated load profiles look realistic both on their own and as part of a wider power systems setting.
Why the authors say this matters
The authors say this matters because low-voltage network operators need more representative loads for planning and congestion management. The study suggests that more realistic synthetic load scenarios could support sub-regional power distribution network planning and operations.
What the researchers tested
The researchers proposed conditional diffusion models for synthesizing daily active and reactive power profiles at the low-voltage distribution substation level. They tested multiple versions, from unconditional generation to generation guided by metadata and daily statistics, and evaluated them using temporal and statistical realism metrics as well as power flow modelling.
What worked and what didn't
The results show that the synthesized load profiles were plausible both individually and as a cohort in a broader power systems context. The Conditional Diffusion model was benchmarked against naive and commonly used generative models and was reported to be effective in producing realistic scenarios. The abstract does not describe which benchmark models performed better or worse in detail.
What to keep in mind
The abstract does not give detailed numerical results or specific performance values. It also does not describe limitations beyond noting that traditional load profiling and earlier generative approaches have struggled to capture substation-level co-behaviour.
Key points
- Conditional diffusion models were used to synthesize daily active and reactive power profiles.
- The generated load profiles were reported to be plausible both individually and as a group.
- Evaluation included temporal realism, statistical realism, and power flow modelling.
- Multiple generation settings were tested, including unconditional and metadata-guided versions.
- The authors say the approach could support sub-regional network planning and operations.
Disclosure
- Research title:
- Conditional diffusion generates plausible low-voltage load profiles
- Authors:
- Alistair Brash, Junyi Lu, Bruce Stephen, Blair Brown, Robert Atkinson, Craig Michie, Fraser MacIntyre, Christos Tachtatzis
- Institutions:
- National Health Service Scotland, Scottish and Southern Energy (United Kingdom), University of Strathclyde, University of Strathclyde, University of Strathclyde, University of Strathclyde, University of Strathclyde, University of Strathclyde
- Publication date:
- 2026-04-20
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.