LF Energy Summit 2024

Centre for Net Zero (CNZ) proposes a hands-on lab-session covering both how to train a generative model to create synthetic smart meter data, and how to evaluate its quality. The contents of the lab will be based on: Faraday, a generative model that we have built and deployed, OpenSynth, an open-source repository of generative models and synthetic smart meter data, and a technical paper on the “Definition of Good” synthetic demand data, co-authored by CNZ and researchers from MIT, Oxford University and Georgia Tech. The aim of the lab is to spread knowledge on how to generate and evaluate synthetic smart meter data to accelerate its adoption across the energy sector, and to promote the OpenSynth project with LF Energy, encouraging more people to join the community. The dataset used will be based on the publicly available Low Carbon London dataset. The algorithms/evaluation could be based on OpenSynth work, or suggested templates/examples that CNZ will prepare. The evaluation of synthetic data will be centred around the core themes of fidelity, utility and privacy. Our target audience for the session is engineers and data scientists with knowledge of training neural networks.