Alliander en Radboud Universiteit werken samen aan energietransitie
Alliander en Radboud Universiteit werken samen aan energietransitie

Improved Pulse Program

AI & Data Science
Duration
1 January 2019
Project member(s)
E.J. Gerritse (Emma) MSc , Jacco Heres (Alliander)
Project type
Research

In the research project Improved Pulse Program, researchers are working on further developing Alliander's Pulse Program. PULSE makes the best possible estimation of transformer station (MSR) loads using smart meter consumption data. This smart meter data may only be used in aggregated and anonymised form, and is only available for a portion of electricity connections. By using methods such as disaggregation and clustering, detailed consumption profiles are created and grouped. These insights help not only to improve daily network management but also to make the power grid future-proof.

What do disaggregation and clustering do?

  • Disaggregation splits the total energy consumption of a connection, measured via the smart meter, into gross generation and consumption. By separating generation and consumption, further analyses, such as clustering users into similar groups, can be performed much more effectively
  • Clustering groups connections with similar consumption profiles. This allows different types of users to be distinguished, such as residential homes, offices or industrial buildings. This helps grid operators to better estimate consumption or generation over time for connections where no smart meter is available, or where smart meter data cannot be used because a time series from an individual customer is needed.

Why is this important? 

By applying disaggregation and clustering, the Improved Pulse Program can provide valuable insights for:

  1. Daily operational planning: Through better predictions of energy flows, grid operators can utilise the power grid more efficiently, avoid peak loads and ensure stability. Meters don't need to be installed everywhere to get a picture of the flows and voltages in the grid.
  2. Network design and planning: The consumption profiles and clusters help with integrating new customers and designing new networks that are tailored to the needs of different types of users. Here too, the major advantage is that, without installing measurements in every transformer house, there is already a reasonably good picture of the energy flows through the transformer.

Results

  • This new approach replaces traditional methods, such as maximum demand indicator measurements where only the maximum current load over a period of many years was recorded. 
  • The project is currently integrating these methods into network operator Alliander's existing PULSE product. PULSE has been live since 2019 and is available for all transformer houses. 
  • The results are also being used in the long-term prediction model ANDES and System Operations' MSR load estimator. 
  • The method and results are also described in the paper (PDF) Creating Bottom Up Load Profiles Using Disaggregation, Clustering and Supervised Machine Learning on Large Smart Meter Dataset.

Funding

Emma and Wieske did their research internship for the Data Science Master's at Alliander and together contributed the equivalent of 6 months full-time to the PULSE program.

Partners

Contact information

More information? Please contact Jacco Heres via jacco.heres[@]alliander.com, or contact: