Adaptive clustering for Decentralised Resilient Energy Management (ADREM)

Joint DST-NWO research project

wind power

The following research groups and people are involved in the project:

Delft University of Technology, Systems Engineering Section
Prof.dr. Frances Brazier (Dutch Principal Investigator, website)
Dr. Martijn Warnier (website)
Selma Čaušević, MSc

Indian Institute of Technology Kanpur, Department of Electrical Engineering
Prof.dr. S.N. Singh (Indian Principal Investigator, website)
K.K. Tomar, Mtech

Centrum Wiskunde & Informatica (CWI), Intelligent Systems Group Han La Poutré (website)
Dr. Michael Kaisers (website)


This project focuses on the design of a framework for Distributed Energy Resource (DER) management based on self-optimizing and self-healing clusters of consumers and producers. Consumers and producers participate in clusters based on negotiated service level agreements (SLAs). Clusters are (approximately) autarkic and adaptive. Cluster membership and SLAs can be (re-)negotiated due to changes in the environment, the (forecasted) availability of energy resources, the overall energy market, but also participants’ forecasts of their own needs and possibilities. This allows for local, decentralised S/D management based on SLAs, reducing complexity on a wider scale; and it provides the basis for stability of the power system through reconfiguration. It thus also allows for preparing load shedding and system restoration / re-configuration schedules for system failures.

The results of the project includes a framework with (1) dynamic profiles and agent models for energy consumers and resource providers together with the technology designed to this purpose; (2) decentralised clustering algorithms as the basis for coordination: criteria, objectives, and boundary conditions; (3) negotiation markets and negotiation strategies for cluster and SLA determination/reconfiguration; and (4) forecasting mechanisms and strategies for different types of loads and markets together with the algorithms and models designed and implemented. The systems are to be evaluated by distributed simulation and emulation during their design, with both simulated and live data.

Adaptation, clustering, decentralized multi-issue negotiation, load profiling, and forecasting thus provide the basis for stability of the power system through optimal reconfiguration and local management of S/D balances.

Project results (publications etc.)



  • Agent based decentralized DC load flow computation
    Kritika Saxena and A.R.Abhyankar
    In the proceedings of the IEEE National Power Systems Conference (NPSC), 2016.
  • Long Term Wind Power Forecast Using Adaptive Wavelet Neural Network
    Bhaskar Kanna and S.N. Singh
    In the proceedings of the 3rd IEEE Uttar Pradesh Section International Conference on Electrical, Computer and Electronics (IEEE UPCON-2016), IIT BHU, Varanasi, 9-11 Dec 2016


Project Events


  • The project organized the International Workshop on Agent Based Self Healing Energy Systems: Design and Implementation, on February 21 - 22, 2017 at IIT Kanpur.
    Slides of all the presentations can be found here.