A scene analysis based operating state prediction method of a power distribution network, comprising the following steps: Step (10): acquiring a network architecture and historical operation information of a power distribution system; Step (20): extracting an output representative scene series segment of a distributed power according to a historical output series of the distributed power; Step (30): obtaining a future single time section T0 multi-scene prediction result by matching similar historical scenes; Step (40): establishing a future multi-time section operating scene tree; Step (50): depth-traversing various scenes in the future multi-time section operating scene tree, performing power distribution network flow analysis for the various scenes, and calculating an out-of-limit risk of a power distribution network line current and an out-of-limit risk of a bus voltage to obtain a future operating state changing trend of the power distribution network including the distributed power. According to the method, the operating state of the power distribution network including the distributed power is predicted, so that the development trend of the operating state of the power distribution network is predicted in real time by means of multi-scene analysis, and risk prediction can be timely implemented.