Dynamic reactive power optimization method for active distribution network with new energy and electric vehicle charging stations connected to the grid

Taking into account new energy output, EV charging load and residential power load, a clustering-local relaxation-correction dynamic reactive power optimization spatio-temporal decoupling strategy is proposed. Firstly, K-medoids clustering algorithm is used to divide the time period. Then, different types of variables are optimized in the same period. Finally, the number of input groups of shunt capacitor banks (SCBS) in several periods is fixed, and the secondary static reactive power optimization is carried out by using continuous variables. According to the characteristics of the model, a hybrid optimization algorithm with cross-feedback mechanism is used to solve different types of variables, and an improved artificial hummingbird algorithm based on Tent chaos mapping and adaptive variation is proposed to improve the solving efficiency. The simulation results show that the proposed decoupling strategy can not only guarantee the dynamic constraints of discrete variables, but also obtain satisfactory optimization results, and the hybrid algorithm can effectively solve the nonlinear optimization problems of mixed integers.