Abstract: To address the supply–demand imbalance and operational uncertainty caused by the integration of distributed energy in smart parks, this paper proposes a availability-driven distributed scheduling optimization method for virtual power plants (VPPs). The method characterizes the stochastic nature of node power and load using a chance-constrained model, and employs cumulants and quantile approximation techniques to transform probabilistic constraints into deterministic forms that can be solved efficiently. On this basis, a utility function integrating availability, power supply quality, and economic cost is constructed, and a two-stage distributed strategy of “neighboring energy borrowing – collaborative refinement” is designed to achieve overload mitigation and load balancing. Case study results show that the proposed method outperforms comparison schemes in terms of service satisfaction, node fairness, and renewable energy utilization, thereby verifying its effectiveness and feasibility for real-time scheduling of park-level VPPs.