Abstract: To address the operational uncertainties and control challenges brought about by the high penetration of renewable energy sources in distribution networks, this paper investigates the collaborative optimization problem of centralized energy storage, intelligent soft-switching (SOP), and demand response (DR). Firstly, a two-layer optimization scheduling model for the distribution network and multiple microgrids was constructed: the upper layer aims to minimize peak shaving and energy storage costs, optimizing the total power of the energy storage system; the lower layer builds upon this, aiming to minimize network losses, voltage deviations, and wind and solar energy curtailment, and coordinating the control of the three-terminal SOP, DR, and traditional reactive power equipment with the energy storage. Secondly, in response to the uncertainty of distributed power generation (DG) output, a conditional generative adversarial network (CGAN) was introduced to generate typical wind and solar power output scenarios. The case study analysis shows: 1) Compared with not participating in the scheduling, incorporating centralized energy storage into the unified dispatch of the distribution network can significantly enhance the stability of operator profits and effectively reduce system network losses and peak shaving costs; 2) The proposed collaborative optimization scheme of the energy storage three-terminal SOP and refined DR model can fully utilize the coordinated control capabilities of active and reactive power, demonstrating significant advantages in reducing network losses, improving voltage quality, and enhancing the utilization of new energy. The strategies proposed in this paper provide an effective approach to improving the economic and safety performance of active distribution networks.