Abstract: As for the complex and diverse spatio-temporal load growth patterns, this paper proposes a dynamic planning method for distribution network based on geographic information system so as to improve the economics and feasibility of distribution network planning. Based on the equivalent location information of load point clustering by fast search and find of density peak, the method uses a line planning method driven by the Bresenham-particle swarm optimization algorithm to form the set of lines to be planned,and uses a mixed integer second-order cone planning model combined with a dynamic planning approach to solve the nonconvex nonlinear problem of distribution network planning. The proposed method simplifies the problem size by modeling the spatiotemporal distribution of load under clustered partitioning, and aims to minimize the comprehensive line cost, including line construction cost, operation and maintenance cost, and loss cost over the whole life cycle. At the same time, the dynamic planning method formulates the planning strategy in phases to solve the structural adjustment brought by the phased network growth, which is suitable for the practical application requirements. Finally, the actual planning area of a city is used as an example to verify that the proposed method can help reduce the construction cost, operation and maintenance cost and improve the planning feasibility.