Classification and Aggregation Method for Load-Based Virtual Power Plants Considering Integrated Response Characteristics

Load resources are usually aggregated in the form of virtual power plants to participate in ancillary service markets, and different ancillary service products have different requirements for the temporal responsiveness of bidding resources. Therefore, when aggregating and managing a large number of controllable loads with significant differences in responsiveness, virtual power plants need to consider not only the power characteristics of controllable loads but also the compatibility of different ancillary service requirements and temporal responsiveness. In this paper, firstly, we proposed a set of peak-shaving and frequency-regulation performance evaluation indicators reflecting the differences in the responsiveness of controllable loads. Then, a two-stage classification method is designed by combining the above indicators with the basic constraints of ancillary services. In addition, a dual-scale distance measure is proposed to measure the similarity of load energy consumption curves, and this measure is used as a feature to implement aggregation of controllable loads by the Ng-Jordan-Weiss spectral clustering algorithm. Finally, the rationality and effectiveness of this method are verified through simulation examples.