Determining Load Profiles for Customers in the Indicated DSO Area based on Smart Meters: A Summary Report for European Data Incubator

LONDON, 17 February 2020 –SMAP Energy was selected to participate in the European Data Incubator – a Horizon 2020 funded program focused on building innovative data applications. This paper details the work undertaken in the project and the results achieved.

Links

 Report Publication – European_Data_Incubator

 Online Demo (to gain access, request login information from info@smapenergy.com) – Link

Authors

Paul Monroe, Riku Arakawa, Yohei Kiguchi, Kristjan Strojan, Alberto Arias

Abstract

The growth in the implementation of smart-meters presents a significant opportunity to improve operational processes in the energy sector. One such opportunity is with distribution system operators using the data to better anticipate the state of the different areas of their networks.

However, large scale smart-meter rollouts often take years to complete, and having digital solutions that can realise outcomes as data becomes increasingly available can improve the return on investment of such rollouts. In this three month project, data from over 2300 metered locations in the Torun region of Poland with varying levels of frequency – ranging from every 15 minutes to readings every few months – are examined to develop a methodology for creating hourly profiles of energy consumption for every location. Numerous patterns in the unavailability of the data are identified, and a methodology is proposed to address each pattern specifically.

The proposed methodology tests multiple machine-learning based approaches – including Random Forest, Decision Tree, Support Vector Regression, and clustering – and assesses their effectiveness both against a generated baseline and other similar studies from academic literature.
Additionally, functional considerations for deploying the algorithmic methodology in a full commercial use case are examined.

The results achieved demonstrate comparable or superior performance against previous studies, leading to the conclusion that the methodology proposed is among the best in the industry, and several areas for future improvement are identified.

Program Information

To read more about our participation in EDI, please visit the website: www.edincubator.eu

Connect with EDI for flash news about our journey:

@EuropeanDataIncubator        @edincubator   EDIncubator H2020