The uncertainty of renewable energy generation challenges the economic and robust operation of microgrid clusters. To address this challenge, this paper proposes a two-stage distributionally robust framework for cooperative scheduling, in which cooperative game theory is used to allocate costs. A Wasserstein distance-based ambiguity set that accounts for external variables is constructed to characterize the uncertainty in photovoltaic (PV) power generation. In addition, a linear decision rule and strong duality are employed to reformulate the distributionally robust optimization (DRO) model into a tractable finite dimensional form. Furthermore, the Shapley value method and the power interaction contribution degree method are used to allocate cooperative costs based on marginal contribution and actual power exchange, respectively. The proposed method is validated on a modified IEEE 33-bus system and is shown to be more economical and less conservative than other methods. Moreover, Monte Carlo simulations show that the DRO-based day-ahead decisions achieve zero load curtailment across all test groups, demonstrating that the proposed method has high operational robustness against PV forecast errors.
