In order to effectively improve the operation safety perception ability of distribution network in the process of ubiquitous power Internet of Things construction, a real-time situation prediction method of distribution network based on multi-time scale state estimation is proposed, which can realize fast and accurate prediction of distribution network security situation. First of all, the multi-timescale recursive dynamic state estimation is realized based on hybrid measurement including distribution PMU through recursive transformation between state projection and pseudo measurement, which realizes the fast equivalent transformation of measurement and shorten the state update period. A correction module is added in the recursive transformation algorithm to eliminate pseudo-measurement fluctuation error, improving the stability and calculation speed of the state estimation algorithm. Then, the real-time safety situation projection of distribution network is realized through partitioned multivariate time series analysis based on historic estimated states, which establishes a real-time state prediction model. Finally, practical examples are simulated and analyzed on MATLAB simulation platform. The simulation results validate the accuracy and effectiveness of the proposed real-time situation prediction method for distribution network.