Abstract: This paper develops a smart distribution network load analysis and forecasting management system based on multivariate data aggregation for the application of multivariate big data in smart distribution networks in order to realize lean management, scientific prediction and rational planning of distribution networks. Firstly, the general framework of the software system is designed; then, each functional module of the software system is developed and introduced; finally, an application example of the software system is given. The system makes full use of the massive historical load data for load characteristic analysis, establishes load characteristic database and industry expansion information database, and realizes load management and maximum load prediction by matching information to new users. In addition, the system establishes a library of load forecasting method model base, which can provide load forecasting functions in different dimensions. From the traditional regional load forecasting to feeder load forecasting, the user access decision is optimized based on current situation of feeder and business expansion information. In general, the system has the features of interoperability of data links between functional modules, information support between different functions, and overall modular design concept, which can meet the daily application requirements of power grid enterprises.