Executive Summary
This evidence is submitted on behalf of a multi-disciplinary University of Southampton team from the EPSRC-funded Programme Grant on Future Electric Vehicle Energy networks supporting Renewables (FEVER) project, which is developing grid-independent EV charging stations, and the Citizen-Centric AI Systems Turing AI Acceleration Fellowship, which investigates building artificial intelligence (AI) systems that can be trusted by all users. The evidence highlights the lack of appliance level energy data as a key barrier to AI enabled applications across residential electricity networks and proposes a privacy preserving, standardised approach for sharing appliance or device level data via smart meters.
Artificial Intelligence (AI) has significant potential to improve the efficiency, affordability, and resilience of the UK energy system. However, realising this potential, particularly in residential and low voltage electricity networks, is currently constrained by the lack of high quality, appliance level energy data. While smart meters provide half hourly aggregate household consumption, they do not offer visibility into the behaviour of individual high impact devices such as Electric Vehicle (EV) chargers, heat pumps, batteries, and electric heating systems.
In the absence of standardised appliance level data, stakeholders rely on either indirect inference, modelling, and estimation approaches - or - on fragmented, proprietary datasets controlled by individual device manufacturers. These approaches introduce uncertainty and increase computational and integration costs, whilst limiting the scalability of AI applications. This, in turn, limits the effectiveness of demand side flexibility, local energy system optimisation and energy management, in addition to low voltage network management, whilst increasing reliance on centralised data collection that raises privacy and security concerns. This submission proposes an open, standardised protocol that allows smart devices to share appliance-level energy data directly with the smart meter. The smart meter would act as a trusted local data hub - generating, storing, and processing data locally by default, with sharing only happening when the user consents. This approach offers several key benefits: it avoids vendor lock-in at both the data and platform level, improves data quality at source, and supports privacy preserving AI by keeping data local and minimising unnecessary export. It also complements existing government initiatives to improve visibility of distributed energy assets.
This evidence suggests the following policy recommendations:
- 1. Support the development of an open, standardised smart device to smart meter data protocol. Government should support an open, vendor neutral protocol that allows appliance level energy data to be shared locally with smart meters, reducing data fragmentation and vendor lock-in while enabling AI enabled energy applications.
-
- 2. Embed privacy by design and consent-based access principles in appliance level energy data governance. Any approach to appliance level energy data should ensure local storage by default, clear limits on data retention, and explicit user consent for data sharing, maintaining trust and protecting consumer privacy while supporting innovation.
Authors
Dr Ezhilarasi Periyathambi, Research Fellow at the University of Southampton working across the FEVER and Citizen-Centric AI Systems (CCAIS) projects.
Fariba Dehghan, researcher at the University of Southampton working at the intersection of AI and sustainable energy.
Professor Andrew Cruden, Professor of Energy Technology at the University of Southampton and Principal Investigator of the FEVER programme grant.
