Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Vickrey-Clarke-Grove based multi agent energy management system

Paper presented at 14th International Conference on Autonomous Agents and Multiagent Systems, AA-MAS 2015, May 4-8, 2015, Istanbul, Turkey
 
: Rohbogner, G.; Maas, M.; Benoit, P.; Wille-Haussmann, B.; Nebel, B.

:
Fulltext urn:nbn:de:0011-n-3790923 (1.4 MByte PDF)
MD5 Fingerprint: a0fbd72ced357f328f3f34afb724f052
Created on: 5.3.2016


2015, 10 pp.
International Conference on Autonomous Agents and Multiagent Systems (AA-MAS) <14, 2015, Istanbul>
English
Presentation, Electronic Publication
Fraunhofer ISE ()
Elektrische Energiesysteme; Systemintegration und Netze - Strom; Wärme; Gas; Elektrische Verteilnetze und Betriebsmittel; systems; VCG; Verteilnetzautomatisierung; Energiemanagement; McAfee

Abstract
Energy management systems (EMS) are being developed to approach challenges emerging with the expansion of renewable energies, electric vehicles, and thermal-electric heating systems. The majority of the developed EMS show a central-hierarchical control structure (e.g. for Virtual Power Plants and grid control centers). In applied research little emphasis has been placed on agent-based EMS and most actors in the energy sector still have reservations towards the application of multi agent systems. Thus, proof of concept studies by means of concrete examples are necessary to show functionality and advantages. By focusing on grid operation and local load flow management in electrical distribution grids, we developed an agent-based EMS applying the Vickrey-Clarke-Groves mechanism. Comparing it against a central-hierarchical system, we show that the agent-based system's mean absolute residual load deviates from the result of the central-hierarchical system by 4.2kW at most, which is 5% of the maximum absolute residual load and 20% of the mean absolute residual load. However, the VCG-Algorithm is the only known approach which ensures both truthfulness in dominant strategies and efficiency, but it lacks when applied in a double-sided trading situation where local electrical demand and local distributed supply need to be matched. With this paper we want to initiate a discourse of how agent-based systems need to be designed for future Smart Grid applications and how they yield benefits comparable to conventional central control architecture.

: http://publica.fraunhofer.de/documents/N-379092.html