The invention discloses a method for obtaining the time-varying substation industry constitution based on the daily load data of the EMS system bottom load outgoing line and the specialized user, which adopts a longitudinal clustering average method to respectively extract the typical daily load curve of each bottom outgoing line. The typical daily load curves of all outgoing lines are analyzed bytransverse clustering, and the clustering results of all outgoing lines under 220kV substation are obtained. Finally, according to the industry attribution of all low-level outgoing lines and dedicated transformer users, the industry composition proportion of the substation is aggregated from bottom to top. The invention is based on the analysis of the EMS system data, and can obtain the composition proportion of the substation industry on-line without manual statistical investigation. It can also obtain the time-varying proportion of the industry, which is conducive to solving the problem oftime-varying load modeling.