Phase-space exploration of unit ensembles in energy management
AbstractCurrently, a transition of the electrical power system occurs that results in replacing large-scale thermal power plants at transmission grid level by small generation units mainly installed in the distribution grid. A shift from the transmission to the distribution grid level and an increase in ancillary service demand is a direct result of this transition, demanding delegation of liabilities to distributed, small energy resources. Decoder-based methods currently are not able to cope with ensembles of individually acting energy resources. Aggregating flexibilities results in folded distributions with unfavorable properties for machine learning decoders. Nevertheless, a combined training set is needed to integrate e. g., a hotel, a small business, or similar with an ensemble of co-generation, heat pump, solar power, or controllable consumers to a single flexibility model. Thus, we improved the training process and use evolution strategies for sampling ensembles.