With the dual carbon construction target of“carbon peak, carbon neutralization”proposed in our country, energy saving and carbon reduction has become a hot issue instantly. User portraits are of great significance for power companies to analyze the power usage behavior of residents and formulate reasonable regulation measures for energy carbon reduction. For this, user carbon portrait label system and portrait method based on power usage behavior are studied. Firstly, collect and filter user multidimensional behavior data. Secondly,according to the filtered data of user’s electricity usage behavior, combined with the purpose of portrait, the label system of user’s carbon portrait is designed from three dimensions：carbon load reduction characteristics of users, low-carbon electricity consumption characteristics of users, carbon production and consumption characteristics of users’electric energy. The data of the sublabels are processed and the multidimensional label data is obtained by analyzing various comprehensive indexes. Finally, by using the k-means clustering algorithm to identify the user’s cluster, and displayed with the three-dimensional scatter plot, the user’s comprehensive carbon reduction index is calculated according to the label system and displayed in a column chart, realizes the visual presentation of the user’s carbon portrait, and reflects the comprehensive ability of the user’s electric energy carbon reduction. The effectiveness of the proposed method is demonstrated by comparing the carbon emissions of typical users before and after the implementation of carbon reduction measures.