16-node loop network encoding results.

For the fault recovery and emergency repair after multiple faults in the distribution network, this paper proposes a fault distribution network recovery strategy considering the collaborative optimization of recovery and emergency repair. Initially, due to the difference and uncertainty between the system load demand and the distributed generation (DG) output, a bilayer dynamic fault recovery with phase type in time scale was constructed. The upper layer considers the recovery of the distribution network during faults, optimizing network reconfiguration schemes using DG outputs predicted by deep stochastic configuration network in conjunction with time-varying load demands. The lower considers the economic effect of the loss during repair and determines the optimal fault repair sequence. Furthermore, an enhanced Nutcracker Optimization algorithm to solve the bilayer model was proposed, determining the dynamic combination of the fault reconfiguration scheme and the repair sequence. Finally, to validate this strategy, this paper conducted simulations using the IEEE 33-node system. The experimental results under multiple strategies show the feasibility and effectiveness of this paper strategy, which ensures the effective recovery of fault nodes after power failure.