Abstract: Comprehensively exploring resources from the existing warehouse network and contractual suppliers to achieve mutual complementarity and collaboration, coordinated inventory and allocation, is a challenge faced by power grid enterprises to reduce cost and increase efficiency in supply chain management, where the selection of optimization models is critical. From the perspective of integrated management, this paper considers demand uncertainty based on current material supply modes in the power grid. It proposes robust optimization models for two types of material supplies based on a deterministic model. A solution algorithm based on the Bender decomposition method is designed to solve the above models. Using data collected from a State Grid case, experiments are designed to analyze the performance of the models and the algorithm. Additionally, the impacts of changes in the penalty cost parameter of unmet demand deviation and the budget uncertainty parameter on total cost are explored under different material supply modes. Results show that 1) the Bender decomposition algorithm in this paper is more efficient in solving large-scale problems. 2) The deterministic model based on expected demand has the lowest total cost but is less effective in addressing uncertain demand risks compared to the two robust models: the robust model based on stochastic scenarios can balance the total cost and the penalty cost of unmet demand deviation, while the robust model based on budget uncertainty can control demand uncertainty within a specific range. 3) Material supply modes considering physical inventory prioritization for allocation and distribution offer better total cost control. 4) The robust model based on budget uncertain sets possesses a good demand satisfaction capability but reveals a greater cost; the robust model based on stochastic scenarios is better than that based on budget uncertain sets in terms of cost, but may cause some unmet demand.