Energy and powerNews

OpenSynth synthetic smart meter data repository launched

OpenSynth synthetic smart meter data repository launched

Image: LF Energy

LF Energy has announced that the OpenSynth model repository for the generation and sharing of synthetic energy demand data is now live.

The OpenSynth initiative, which was originated by the Octopus Energy Group’s Centre for Net Zero and is now open sourced under LF Energy, is aimed to overcome the challenge of sharing smart meter data due to privacy protections.

Smart meter data is essential for understanding and adapting to changing demand profiles, with its granularity and real time nature able to capture the dynamic nature of energy flows with the rise of variable renewable energy sources and behind-the-meter technologies such as heat pumps, electric vehicles and battery storage.

Moreover, generation of synthetic data is considered the fastest way to achieve widespread access to demand datasets, rather than challenging current data regulations and smart meter legislation.

Have you read?
Smart meter data innovations to trial in Britain
The good, the bad and the safety: AI and data in the energy sector

The goals of the OpenSynth initiative are to create a one-stop-shop for synthetic energy data, to encourage more people to create and share synthetic data and to spread its adoption in academia, industry and within government in order to accelerate the decarbonisation of energy systems.

Its basis is over 300 million smart meter data readings from Power Networks, with associated metadata such as property type and low carbon technology ownership.

These data have been used to train a generative AI model named ‘Faraday’, which generates daily load ‘profiles’ consisting of half-hourly kWh consumption for a given set of user-specified inputs, e.g. low carbon technology, property type and/or season, and outputs synthetic data that can be shared with third parties.

A common evaluation framework to benchmark the performance of various synthetic data generation algorithms is also proposed.

Looking ahead the next developments that are planned for OpenSynth are the release of synthetic data generated with the Faraday model, trained on Octopus Energy datasets, into the data repository and the implementation of the evaluation framework.

In a statement, LF Energy describes the go-live of OpenSynth as “a significant milestone for the energy industry and the global community of researchers, modellers and policymakers dedicated to advancing energy transitions.

“OpenSynth is poised to revolutionise the way we access and utilise energy demand data. By building a community of data holders and innovators, we aim to democratise access to energy demand data and drive the global transition to a more sustainable energy future.”

The OpenSynth model repository is available on github and PyPI.