With the emergence of power load diversification, the traditional visualization technology is difficult to effectively analyze and display the load section data of the power grid. In this paper, a visualization method is proposed to show the sectional load of distribution network in the form of bubble diagram with multi-attribute parameters, which further strengthens the analysis of the distribution law of load types. Firstly, qualitative analysis of visual data objects are conducted, and the outliers of section load are cleaned by standard deviation method, and the load information and geographic information after cleaning are weighted by PCA method. Then, the weighted results are input into the MiniBatch-Kmeans algorithm for clustering to obtain the category labels of the load points. Furthermore, in order to visualize the operation situation of distribution network, the information of load, geography, time and ID is displayed as dynamic bubble diagram in the example and the traditional load heat map is compared. The results show that the visualization method proposed in this paper can reflect more information of operation status of power grid, which presents the multi-parameter data to reflect the distribution law and development law of load, and can improve the perception ability of dispatchers to the operation data of distribution network.