Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Faraday: Synthetic Smart Meter Generator for the smart grid

Sheng Chai, Gus Chadney

TL;DR

Faraday tackles the privacy barrier to granular smart-meter data by introducing a conditional variational autoencoder with a Gaussian mixture latent space to synthesize household-level load profiles conditioned on low-carbon technology ownership. The approach uses Maximum Mean Discrepancy loss and quantile losses to improve fidelity across distributions, and samples from a learned GMM to generate realistic time series; utility is validated through a Train-on-Synthetic, Test-on-Real evaluation and alignment with substation data. Privacy safeguards include k-anonymity, partial black-box access, and limiting outputs to daily profiles, with a need for explicit quantitative privacy assessment. The work demonstrates practical potential for grid modelling and digital twins, while outlining avenues for stronger privacy guarantees and longer-horizon outputs.

Abstract

Access to smart meter data is essential to rapid and successful transitions to electrified grids, underpinned by flexibility delivered by low carbon technologies, such as electric vehicles (EV) and heat pumps, and powered by renewable energy. Yet little of this data is available for research and modelling purposes due consumer privacy protections. Whilst many are calling for raw datasets to be unlocked through regulatory changes, we believe this approach will take too long. Synthetic data addresses these challenges directly by overcoming privacy issues. In this paper, we present Faraday, a Variational Auto-encoder (VAE)-based model trained over 300 million smart meter data readings from an energy supplier in the UK, with information such as property type and low carbon technologies (LCTs) ownership. The model produces household-level synthetic load profiles conditioned on these labels, and we compare its outputs against actual substation readings to show how the model can be used for real-world applications by grid modellers interested in modelling energy grids of the future.

Faraday: Synthetic Smart Meter Generator for the smart grid

TL;DR

Faraday tackles the privacy barrier to granular smart-meter data by introducing a conditional variational autoencoder with a Gaussian mixture latent space to synthesize household-level load profiles conditioned on low-carbon technology ownership. The approach uses Maximum Mean Discrepancy loss and quantile losses to improve fidelity across distributions, and samples from a learned GMM to generate realistic time series; utility is validated through a Train-on-Synthetic, Test-on-Real evaluation and alignment with substation data. Privacy safeguards include k-anonymity, partial black-box access, and limiting outputs to daily profiles, with a need for explicit quantitative privacy assessment. The work demonstrates practical potential for grid modelling and digital twins, while outlining avenues for stronger privacy guarantees and longer-horizon outputs.

Abstract

Access to smart meter data is essential to rapid and successful transitions to electrified grids, underpinned by flexibility delivered by low carbon technologies, such as electric vehicles (EV) and heat pumps, and powered by renewable energy. Yet little of this data is available for research and modelling purposes due consumer privacy protections. Whilst many are calling for raw datasets to be unlocked through regulatory changes, we believe this approach will take too long. Synthetic data addresses these challenges directly by overcoming privacy issues. In this paper, we present Faraday, a Variational Auto-encoder (VAE)-based model trained over 300 million smart meter data readings from an energy supplier in the UK, with information such as property type and low carbon technologies (LCTs) ownership. The model produces household-level synthetic load profiles conditioned on these labels, and we compare its outputs against actual substation readings to show how the model can be used for real-world applications by grid modellers interested in modelling energy grids of the future.
Paper Structure (11 sections, 5 figures)

This paper contains 11 sections, 5 figures.

Table of Contents

  1. Introduction
  2. Methodology
  3. About the dataset
  4. Faraday architecture
  5. Modifications to VAE
  6. Gaussian sampling of latent space
  7. Results
  8. Fidelity
  9. Utility
  10. Privacy
  11. Future work and Conclusion

Figures (5)

  • Figure 1: Faraday architecture
  • Figure 2: Faraday outputs at various quantiles. Y-axis is the kWh consumption. X-axis is the half-hourly periods where 1 is 00:00hrs and 48 is 23:30 hrs.
  • Figure 3: Faraday outputs at various quantiles conditioned by whether households own an electric vehicle (EV).
  • Figure 4: Fidelity of Faraday outputs.
  • Figure 5: Utility of Faraday outputs.