With the increasing of frequency of data acquisition and the dimension of data analysis, the analysis of household electricity consumption behavior becomes more complex. The development of user tag and portrait technology brings a more intuitive and concise expression to the analysis of user's electricity consumption. Based on massive user archives, power load and electricity consumption data, considering the user's power consumption characteristics and influencing factors, a user behavior tag library is built. Fuzzy clustering algorithm is used to cluster tags and achieve different types of electricity consumption behavior. The results of 2000 industrial and commercial users show that the selected tags of electricity consumption behavior is reasonable, the clustering algorithm is effective, and user portraits is precise. The results can provide powerful support for power companies to understand the user's electricity consumption habits, mine user's electric power needs and improve service level.