Research on data-driven distributionally robust optimization strategies for new energy systems incorporating energy storage

Aiming at the issues of uncertainty in new energy output, low power quality, and insufficient system stability in distributed new energy systems incorporating energy storage, this paper proposes a data-driven two-stage distributionally robust optimization strategy. Different from traditional robust optimization models, this strategy adopts a two-stage approach. Furthermore, unlike conventional uncertainty set construction methods, this paper constructs a Wasserstein uncertainty set to characterize uncertainty from the perspective of empirical distribution using a data-driven approach, where the probability mass of the distributions of uncertain variables is higher. In model processing, the strong duality principle is applied to simplify the problem, McCormick envelopes and the big-M method are used to handle bilinear terms, and the improved Benders decomposition method is employed to solve the model, yielding the optimal dispatch solution. The proposed method is validated through case analysis using real-world data from both the distributed new energy network system and the Institute of Electrical and Electronics Engineers 18-bus system, demonstrating its effectiveness.