With the large-scale development of new elements such as distributed new energy,distributed generation and electric vehicles under the new power system,the distribution network voltage control is faced with challenges such as heterogeneous control object characteristics,complex models and variable scenarios. In this context,a distribution network voltage control strategy based on model-free adaptive control is proposed. The data from dominant node voltages,distributed generation,and energy storage device output are used by this strategy to establish a real-time dynamic linearized data rolling pool based on the distribution network voltage control partitioning and dominant node selection. Based on the characteristic model theory and data drive,the complex system voltage control characteristics are time-varyingly modified by the controller characteristic parameters,and the control model complexity is reduced without losing system model information. Finally,the effectiveness of the proposed control strategy is shown in the results by simulating and comparing in the modified IEEE 33-node distribution network system arithmetic. The response of the proposed control strategy is quicker than that of traditional model-based control method,thus reducing the voltage fluctuation caused by disturbances. The proposed control strategy has good application prospects.