Chris Briggs
School of Computing and Mathematics
Reducing Carbon Emissions by Enabling Consumer Acceptance of Smart Meters Through a Novel Privacy-preserving Energy Forecasting Application
The UK ’ s smart energy meter rollout has been plagued with complications . Acceptance of smart meters has been eroded by news of older meters becoming obsolete when switching supplier , and a lack of evidence that the meters have a positive effect on reducing consumer energy consumption . Additionally , consumers cite data privacy concerns as one of the major reasons for not accepting a smart meter installation . Smart Energy GB , the organisation charged with leading the " Campaign for a smarter Britain ", recently evolved its advertising messaging from ' Helping you save energy in the home ' to ' Doing your bit to help upgrade the UK ' s energy system '. To this end , accurate energy consumption forecasting is critical to provide insights to the energy sector ( e . g . for planning , optimisation etc ) that may enable drastic reductions in carbon emissions . This recent work outlines how federated learning can be applied to energy consumption forecasting whilst protecting consumer ’ s private data . This would allow a machine learning forecasting application to run on smart meters and prevent sharing of raw high-resolution energy consumption data which might otherwise reveal highly sensitive behavioural patterns of a household ' s occupants . This research is undertaken as part of the Smart Energy Network Demonstrator ( SEND ) at Keele University .
Postgraduate Conference 2020 Page 14