To address the issue of insufficient transient reactive power support in receiving-end power grids due to the widespread application of high-voltage DC transmission and new energy generation, a two-phase reactive power dynamic partitioning method considering both speed and adaptability is proposed. First, transient voltage characteristics of nodes are constructed based on the margin between transient voltage and its threshold, as well as the steady-state voltage recovery level after a fault. Using these characteristics, electrical distances between nodes are derived. Then, a two-stage partitioning strategy is designed. In the first stage, initial partitioning of reactive power sources and loads is realized through reactive voltage sensitivity analysis; in the second stage, a hierarchical clustering algorithm is used to merge and optimize the initial region in combination with the transient voltage characteristics. The optimal scheme is selected using the modularity index, while reactive margin and coupling degree indexes are applied to validate the rationality of the partitioning. Second, dynamic adjustment criteria are designed based on the transient voltage characteristics under different typical scenarios to realize adaptive optimization of the partition structure. Finally, simulation results based on an improved New England 39-bus test system verify the effectiveness of the proposed method