Abstract: Urban energy systems not only need to have high reliability for small-scale and large-probability daily failure events, but also need to have high resilience to cope with large-scale and small-probability extreme disaster events. With the increasing coupling of cross-energy forms and power electronization, the urban energy system is expected to improve system resilence through coordinated cross-system energy transfer and rapid regulation of massive power electronic devices. First, this paper proposes a twostage resilience operation model for urban energy systems under extreme disaster events, which integrates the modeling and coordination of fast dynamic characteristics of power electronic devices and slow dynamic characteristics of multi-energy networks.Then, to address the problems of multiple constraints and high solution complexity of the resilience model, a data-driven urban energy system security rule extraction method based on space-weighted inclined decision tree theory is proposed. The method transforms the complex resilience model into a small number of mixed-integer linear constraints, namely security rules, and embeds them into the system normal operation model. Finally, the effectiveness of the proposed method and its advantages in computational efficiency over the traditional model-driven methods are verified by case analysis.