Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
Two-stage robust optimization dispatch for multiple microgrids with electric vehicle loads based on a novel data-driven uncertainty set
Ist Teil von
International journal of electrical power & energy systems, 2022-01, Vol.134, p.107359, Article 107359
Ort / Verlag
Elsevier Ltd
Erscheinungsjahr
2022
Link zum Volltext
Quelle
Alma/SFX Local Collection
Beschreibungen/Notizen
•A two-stage robust scheduling model is proposed for multiple microgrid systems with electric vehicle loads.•A new cluster charging model is proposed for electric vehicle aggregators.•An improved data-driven uncertainty set is obtained using a neural network framework.•The optimization problem is solved by an improved data-driven column and constraint generation algorithm.
The uncertainty in electric power demand associated with the increasing penetration of electric vehicles can profoundly affect the stability of microgrids. Therefore, the present work addresses the need to reduce the operating cost of multi-microgrids and improve the convergence performance of the solution algorithms applied for their optimized electric power dispatch when considering the uncertainties associated with existing loads, renewable energy sources, and electric vehicle usage by proposing a novel double-layer robust optimization dispatch model. The multi-microgrid layer updates the constraints according to the number of real-time electric vehicle insertions and controls multi-type energy units to obtain the lowest operating cost under worst-case scenarios. The electric vehicle aggregator layer achieves the smallest power fluctuations and most ideal charging processes by controlling charging power while avoiding safety violations. Unrealistic worst-case scenarios are eliminated in an improved data-driven uncertainty set that is obtained from a neural network trained using a large volume of historical data. The proposed approach accurately captures the characteristics of the system, and thereby enables the replacement of a large volume of historical data with characteristic system features when formulating the uncertainty set, which simultaneously maintains high reliability and significantly reduces the convergence time. The optimization problem is then transformed into a mixed integer linear programming problem and finally solved by a column and constraint generation algorithm. The feasibility of the proposed approach and its superiority in economy and convergence performance compared to existing robust optimization methods are verified by numerical case studies.