Optimal Scheduling of Electric Vehicle Virtual Power Plants Considering Source-load Correlation

The construction of a new power system and the increasing share of new energy sources have brought challenges to power and electricity balance. Virtual power plants are an effective way to address the optimal operation issues of distributed generation systems. Considering the economy benefits of electric vehicle (EV) demand response and the uncertainties related to system sources and loads, this study proposes a day-ahead optimal scheduling model for EV virtual power plants incorporating a carbon trading mechanism. Firstly, the improved diffusion kernel density estimation is applied to fit the data of wind and solar power generation, conventional loads, and EV electricity loads. A joint probability distribution model considering the spatio-temporal correlation of sources and loads is constructed using the R-vine Copula model, and source-load correlated data scenarios are generated. Subsequently, an EV demand response mechanism is defined according to the characteristics of EV electricity loads, and a stepped carbon trading mechanism is introduced to build a carbon emission cost model. Finally, according to the interactive operation between the EV virtual power plant and the power grid, the day-ahead optimal scheduling is carried out using its internal distributed energy resources. Case studies demonstrate that the joint distribution scenarios considering source-load correlation exhibit a higher degree of matching. Compared with conventional virtual power plants, the proposed dispatch model for electric vehicle virtual power plants demonstrates significant optimization effects in terms of system operation cost reduction, carbon emission mitigation, and comprehensive cost-effectiveness enhancement.