Modern intelligent distribution network can achieve rapid power recovery by dispatching its “flexible sources” of the demand side. This paper proposes a fast power restoration strategy based on network equivalence and parallelization aiming at the characteristics of intelligent distribution network which can use power network resources flexibly to cope with different electrical energy loss. The algorithm uses special power supply nodes (called equivalent sources) to replace surrounding feeders, adopts parallel computation mode of combining off-line equivalent sources pre-storage with real-time ones. The off-line data is pre-stored and preferred to use. While facing larger electrical energy loss, the algorithm calculates in parallel the real-time equivalent sources corresponding to different dispatching possibilities. In order to further improve the efficiency, the algorithm uses improved adjacency matrix to maintain the radial topology and reconstruct the network efficiently. Several cases of different electrical energy loss are studied to show that this strategy can achieve rapid power recovery by using different support sources reasonably and efficiently. This work is supported by a Science and Technology Project of State Grid Zhejiang Electric Power Company (No. 5211JY16000X).