Skip to main content
placeholder image

A dynamic evolutionary strategy for time ahead energy storage management in microgrid

Conference Paper


Abstract


  • Based on a reliable prediction method, this paper proposes a dynamic evolutionary frame in order to maximize the economic benefits for micro-grid (MG) over time. The optimal solution is a series of time related control vectors for the whole time span, which can control the energy storage to charge and discharge optimally by the appropriate energy quantity and at the appropriate time instants. The main research objectives are not only to optimize the energy utilization of renewable sources, e.g. wind farm or solar PVs integrated with storage devices given the energy demand and dynamic electricity pricing at any time instant, but also to consider optimal planning in global angle. Extreme learning machine (ELM) method is used to provide the next state prediction at each time slice, and a constrained evolutionary strategy is used to explore the local optimal solution within each time slice. Furthermore, to take into account of the forecasting errors, 5 independent predictions are processed in parallel. Finally, the proposed approach has been verified not only on its forecasting aspect, but also on solving deterministic benchmarks which are useful for complex dynamics. Experimental results demonstrate the promising performance of the proposed method.

Publication Date


  • 2016

Citation


  • Xiao, C., Soetanto, D., Muttaqi, K. & Zhang, M. (2016). A dynamic evolutionary strategy for time ahead energy storage management in microgrid. 2016 IEEE International Conference on Power System Technology, POWERCON 2016 (pp. 1-6). IEEE Xplore: IEEE.

Scopus Eid


  • 2-s2.0-85006818115

Ro Metadata Url


  • http://ro.uow.edu.au/eispapers/6493

Start Page


  • 1

End Page


  • 6

Abstract


  • Based on a reliable prediction method, this paper proposes a dynamic evolutionary frame in order to maximize the economic benefits for micro-grid (MG) over time. The optimal solution is a series of time related control vectors for the whole time span, which can control the energy storage to charge and discharge optimally by the appropriate energy quantity and at the appropriate time instants. The main research objectives are not only to optimize the energy utilization of renewable sources, e.g. wind farm or solar PVs integrated with storage devices given the energy demand and dynamic electricity pricing at any time instant, but also to consider optimal planning in global angle. Extreme learning machine (ELM) method is used to provide the next state prediction at each time slice, and a constrained evolutionary strategy is used to explore the local optimal solution within each time slice. Furthermore, to take into account of the forecasting errors, 5 independent predictions are processed in parallel. Finally, the proposed approach has been verified not only on its forecasting aspect, but also on solving deterministic benchmarks which are useful for complex dynamics. Experimental results demonstrate the promising performance of the proposed method.

Publication Date


  • 2016

Citation


  • Xiao, C., Soetanto, D., Muttaqi, K. & Zhang, M. (2016). A dynamic evolutionary strategy for time ahead energy storage management in microgrid. 2016 IEEE International Conference on Power System Technology, POWERCON 2016 (pp. 1-6). IEEE Xplore: IEEE.

Scopus Eid


  • 2-s2.0-85006818115

Ro Metadata Url


  • http://ro.uow.edu.au/eispapers/6493

Start Page


  • 1

End Page


  • 6