Abstract: In-depth exploration of user load characteristics and adjustable potential are in urgent need for the refined development of the power market under the background of power big data. This paper proposed a comprehensive classification method of user power consumption behavior considering user load characteristics and adjustable potential, which was suitable for the situation where the amount of power system load data was large and the user power consumption behavior had many influence factors. First, through the multi-dimensional analysis of the influencing factors of users' electricity consumption behavior oriented to electric power big data, an implementation framework for comprehensive analysis of electricity consumption behavior considering the characteristics of user load and adjustable potential was proposed. Secondly, in order to realize accurate clustering under the influence of multi-dimensional influencing factors of users' electricity consumption behavior, this paper designed a comprehensive clustering method that combined K-means and SOM for secondary clustering and BP neural network for reverse adjustment and correction. Finally, the effectiveness and practicality of the classification method proposed in this paper were both verified by selecting the measured load data and user electricity consumption behavior related influencing factor data in Ireland. The method also proved the generalization ability of the method for multiple scenarios.