Abstract: In view of the uncertainty caused by distributed generation and load fluctuation in the operation of active distribution network, the development and utilization of distribution data resources and value mining are further deepened, and a data-driven framework is proposed to define the active control method of distribution network. Firstly, based on the time series voltage and load data, the time series sampling function of each node is generated, and various grid components are configured to establish the topological relationship. On this basis, combined with multi-dimensional operation index constraints, the multi period voltage dynamic optimization model of active distribution network with distributed generation is established with the objective function of minimizing conventional peak load and meeting the requirements of multi period voltage fluctuation. At the same time, the mixed integer second-order cone programming method is combined to simplify the optimization model, so that the optimization problem is transformed into a linear model, and the Gauss Seidel method is used for iterative solution to obtain the optimization scheme, so as to promote the effective consumption of distributed generation in distribution network and the safe and stable operation of distribution network. The effectiveness of the model and control method is verified on the improved ieee33 node example, and the dynamic matching between the voltage regulation algorithm and the consumption capacity of the power system is investigated. The analysis shows that the proposed data-driven algorithm can effectively improve the utilization rate of new energy, and has a good supporting effect on the economy of distribution network.