In order to solve the problems of increasing fluctuation of distributed generation, deepening coupling of grid-load and response lag of traditional centralized scheduling in digital grid, this paper proposes a cloud-edge collaborative scheduling strategy for “source-grid-load-storage” intelligent fusion terminal. In this paper, multi-source perception, edge computing, rolling optimization and closed-loop feedback are unified into the same scheduling framework. The terminal completes state acquisition, local recognition and rapid control, and the cloud is responsible for global prediction, linkage solution and policy update. Experimental verification was carried out in typical park distribution scenarios. The results show that compared with the traditional centralized dispatching, the proposed method reduces the average daily operating cost of the system by 14.11%, the average deviation between supply and demand by 57.14%, the average dispatching response delay is shortened from 42 s to 17 s, and the new energy consumption rate is increased from 84.6% to 92.1%. The strategy shows good adaptability to high-frequency disturbance and local mutation in typical campus level scenarios, and can provide a reference computing framework for intelligent scheduling at the terminal side of digital power grid.
