The spatial distribution of urban power load provides accurate data of load size and its spatial location

which is the basis and precondition for the evaluation of current distribution network and spatial load forecasting. A clustering analysis method for urban load spatial distribution is proposed. Open source information of users in planning areas on Baidu map based on Python crawler technology is collected and then the building where the users are located and their attributes is identified by regular matching function. Power load is estimated according to the power consumption per unit area of building body to construct a sample set of spatial distribution of power load in which the samples have three attributes: time

space and load power. Further

clustering analysis of medium and low voltage user load is performed based on the two indexes of sample local density and sample spacing and the coordinates of load center

the local load density and the distribution radius of load clusters are calculated on the basis of the clusters' attributes. Finally

taking a power grid of a city as an example

the consistency of the clusters' attribute indicators is compared with the planning data. Rational analysis of substation configuration is carried out and the accuracy and effectiveness of the proposed method is illustrated.