Distributed Optimization of Energy Costs in Manufacturing using Multi-Agent System Technology

Abstract

While widely endorsed, the increased provision of electricity from renewable sources comes with the concern that energy supply will not be as reliable in the future as it is today, due to variations in the availability of wind and solar power. However, fluctuations in energy supply also give rise to volatility of the price for short-term energy procurement, and therefore bear the opportunity to save costs through shifting energy consumption to periods of low market prices. In a previous work, we presented an evolution-strategy-based optimization of production schedules with respect to day-ahead energy price predictions, yielding good results, but -- being a stochastic optimization -- not always arriving at the best solution. In this paper, we extend our framework by agent-based mechanisms for distribution and parallelization of the optimization, to increase scalability and reliability of the approach.

@INPROCEEDINGS{Kuester2012Distributed,
  author = {Tobias K{"u}ster and Marco L{"u}tzenberger and Daniel Freund},
  title = {Distributed Optimization of Energy Costs in Manufacturing using Multi-Agent
	System Technology},
  booktitle = {Proceedings of the 2textsuperscript{nd} International Conference
	on Smart Grids, Green Communications and IT Energy-aware Technologies,
	Maho Beach, St. Maarten},
  year = {2012},
  editor = {Pascal Lorenz and Kendall Nygard},
  pages = {53--59},
  month = {March},
  publisher = {IARIA},
  abstract = {While widely endorsed, the increased provision of electricity from
	renewable sources comes with the concern that energy supply will
	not be as reliable in the future as it is today, due to variations
	in the availability of wind and solar power. However, fluctuations
	in energy supply also give rise to volatility of the price for short-term
	energy procurement, and therefore bear the opportunity to save costs
	through shifting energy consumption to periods of low market prices.
	In a previous work, we presented an evolution-strategy-based optimization
	of production schedules with respect to day-ahead energy price predictions,
	yielding good results, but - being a stochastic optimization - not
	always arriving at the best solution. In this paper, we extend our
	framework by agent-based mechanisms for distribution and parallelization
	of the optimization, to increase scalability and reliability of the
	approach.}
}
Autoren:
Kategorie:
Tagungsbeitrag
Jahr:
2012
Ort:
Proceedings of the 2nd International Conference on Smart Grids, Green Communications and IT Energy-aware Technologies, St. Maarten, St. Maarten, pp. 53-59 (Winner of the Best Paper Award)