This paper addresses the multi-objective optimization problem of grid scheduling under multi-terminal information interaction architecture, and proposes a grid scheduling optimization model based on adaptive dynamic planning under grid-connected mode. Taking operation cost and environmental cost as the core objectives, the multi-objective optimization scheduling model of microgrid under grid-connected mode is constructed. The idea of adaptive dynamic planning is introduced, and an improved iterative ADP algorithm is designed by combining neural networks. The model is verified by examples to generate a scheduling scheme that takes into account both economy and environmental protection: in 24-hour scheduling, the discounted solution operating cost is RMB 395.6, the environmental cost is RMB 216.0, and the storage charging and discharging strategy interacts with the main grid to dynamically match the loads and the fluctuation of electricity price. Comparative analysis shows that the iterative ADP algorithm reaches the optimal value at 250, 440, and 150 iterations, respectively, and the scheduling results satisfy the power balance and unit operation constraints while outperforming the traditional ADP algorithm.
