According to the user load characteristics and the controllable load time-domain response features and the regulation potential of auxiliary services in virtual energy storage plants(VSP)

this paper proposes a peak shaving performance index

which can reflect the difference in response performance of flexible resources. Based on the index

a day ahead optimization model aiming at maximizing the operating income of VSP and a time-ahead rolling and real-time correction optimization model aiming at minimizing the penalty cost of VSP output changes in rolling time domain are established respectively. In the day ahead stage

the linear programming algorithm is used to solve the day ahead optimal declared power and the number of peak shaving contracts signed with users. In the intra-day stage

the rolling optimization strategy is introduced into the composite differential evolution algorithm to solve the model

forming a rolling tracking and correction feedback for the output deviation in the intra-day and time-ahead/real-time stages. Finally

an example is given to verify the advantages of the proposed model and strategy in improving the flexibility of demand side resource peak shaving and enhancing the capacity of VSP emergency peak shaving.