Abstract: Under the situation of building the new power system, the geographic characteristics of multiple natural disasters in China make the urgency and complexity of disaster prevention and resilience increasing. To enhance the resilience of the distribution network against disasters, it is essential to consider the impact of disasters during the planning phase. This paper presents a distribution network bi-level planning method for disaster prevention considering distributed generation collaborative configuration. Firstly, a probabilistic model for line failures under different disaster conditions is introduced for studying load outage. Then, a bi-level planning model is established. The upper level determines the location and scale of distributed generators, while the grid structure is determined using the Steiner minimum tree algorithm with the objective of minimizing the system costs. The lower level focuses on minimizing outage costs while adhering to the safety constraints of power system operation. An improved NSGA-II algorithm and simulated annealing algorithm are employed to solve the proposed model. Finally, simulation experiments are conducted on a 20-bus system to validate the effectiveness of the proposed approach.