Abstract:This paper aimed on providing positive user service by big data analyzing and data mining. According to the document No.1 published by StateGrid on 2018, user experiences are putting on hold through excellent service delivery. Taking use of power users’ load data through electricity information acquisition system, a set of models is constructed to recognize user side fault, like anomaly analysis model of distribution line and metering device based on voltage data and anomaly analysis model of distribution line based on station impedance and theoretical line loss. These models help us find problems of electric energy meter abnormal, unbalanced three-phase load, wrong relationship between users and transformers. According to the study, we make some achievements on discovering potential faults, realizing positive judgment of low-voltage fault, reducing fault outage, accurately locating fault points and reducing workload of on-site inspection. The results of study not only meet user’s demand of no power outage and reducing power outage, but also realize fault fast analysis and accurate location at smart grid user side, Through this study, we can improve power supply capacity of urban and rural distribution networks and make effort on keeping higher reliability on power distribution system.