Aiming at the load fluctuation problem caused by a high proportion of new energy grid connections, a reactive power optimization method based on deep reinforcement learning (DRL) considering topological characteristics is proposed. The proposed method transforms the reactive power optimization problem into a Markov decision process and models and solves it through the deep reinforcement learning framework. The Dueling Double Deep Q-Network (D3QN) algorithm is adopted to improve the accuracy and efficiency of calculation. Aiming at the problem that deep reinforcement learning algorithms are difficult to simulate the topological characteristics of power flow, the Graph Convolutional Dueling Double Deep Q-Network (GCD3QN) algorithm is proposed. The graph convolutional neural network (GCN) is integrated into the D3QN model, and the information aggregation of topological nodes is realized through the graph convolution operator, which solves the calculation problem of deep learning algorithms in non-European space and improves the accuracy of reactive power optimization. The IEEE standard node system is used for simulation analysis, and the effectiveness of the proposed method is verified.

To further investigate the complementary characteristics among subsystems of the combined electricity–gas–heat system (CEGHS) and to enhance the renewable energy accommodation capability, this study proposes a comprehensive optimization scheduling framework. First, an optimization model is developed with the objective of minimizing the total system cost, incorporating key coupling components such as combined heat and power units, gas turbines, and power-to-gas (P2G) facilities. Second, to address the limitations of traditional robust optimization in managing wind power uncertainty, a distributionally robust optimization scheduling model based on Hausdorff distance is constructed, employing a data-driven uncertainty set to accurately characterize wind power fluctuations. Furthermore, to tackle the computational challenges posed by complex nonlinear equations within the model, various linearization techniques are applied, and a two-stage distributionally robust optimization approach is introduced to enhance solution efficiency. Simulation studies on an improved CEGHS system validate the feasibility and effectiveness of the proposed model, demonstrating significant improvements in both economic performance and system robustness compared to conventional methods.