Abstract: Distribution network planning of distributed generation has to analyze the characteristics of voltage profile at different nodes before and after the equipment access. Due to the large number of nodes and the stochastic fluctuation of both loads and distributed generation, the efficiency and accuracy assessment method has still been a research hotspot. This paper proposed a multi-dimensional typical scenarios generation algorithm for distribution network based on load clustering and network equivalence. First, based on vector K-medoids clustering method, load nodes were divided into n types according to power fluctuation characteristics. Besides, distributed generation nodes were divided into photovoltaic and wind turbine. Second, a method was presented to calculate the weight coefficient of nodes, representing the influence on line voltage drop. Consequently, a radial distribution network was simplified to n + 2 single-node equivalence. Third, n + 1 D and 1D K-means clustering methods were applied to compress the power data of load and that of photovoltaic and wind generation respectively. Then, typical power scenarios of the initial network are calculated according to the clustering center. Finally, a verification with an example of a practical distribution network showed the feasibility of this proposed algorithm and the accuracy of voltage profile assessment in typical scenarios.