Abstract: For the nonlinear optimization problem of coordinated scheduling of discrete devices and continuous variables in multi-source distribution networks, this paper develops a double-layer coordinated optimization scheme of active and reactive power for distribution networks, which combines the improved bald eagle search(BES) algorithm with second-order cone programming (SOCP). First, a multi-objective active and reactive power coordinated optimization model is established by comprehensively considering the costs of electricity purchasing, operation and maintenance, grid loss, and wind curtailment. Second, by introducing the Halton sequence and refractive opposition-based learning to improve the population initialization, and embedding the golden sine operator and dynamic spiral search mechanism in the dive stage, the global exploration ability and local convergence accuracy of the BES algorithm are enhanced. Finally, to address the difficulty in solving the coupling of discrete and continuous variables, a double-layer optimization architecture is established. The upper-layer improved BES conducts heuristic search for the discrete device states, and the lower-layer SOCP is nested to achieve the exact solution for the convex relaxation of continuous variables. Simulation results comprehensively verify the effectiveness and superiority of the proposed strategy.

Key words: distribution network, active-reactive power coordination, two-layer optimization, improved bald eagle search(BES), second-order cone programming(SOCP)