With the construction and development of distribution network, the penetration rate of distributed generation (DG) represented by photovoltaics is gradually increasing. It becomes possible that DGs provide short-term support for power supply restoration in distribution network. However, the randomness of DG output brings new challenges to the reliability of power supply restoration, such as the possibility of secondary load shedding in the restored area. On this basis, this paper takes photovoltaic as an example and uses time variation to measure randomness, and proposes a short-term power supply restoration strategy for the distribution network considering the time variation of DG output. Firstly, a power supply restoration model for the distribution network with SOP is established. Then, based on the power supply restoration model, the principle of the power supply restoration strategy for distribution network based on the forecast of photovoltaic output is introduced, and the decision-making process of the power supply restoration plan is elaborated. Finally, the improved IEEE33 node model is taken as an example to verify the proposed restoration strategy, which can reduce the time-variation effect of the photovoltaic output on the reliability of the power supply restoration. The results show that the proposed strategy can improve the call speed of the recovery plan after a failure, and it has positive significance for improving the reliability of the power supply recovery of distribution network.