Abstract: In the aftermath of extreme disasters, electric vehicles（EV） can serve as mobile power sources, bolstering the resilience of distribution networks by supplying power to critical loads. To address the challenge of dispatching instruction instability stemming from unforeseeable faults, alongside the need for seamless switching between normal and fault state scheduling models in distribution networks, this study explores the integration of EVs into resilience enhancement strategies, with smooth switching of the running state. Based on the normalization of binary variables representing the EV state and connection mode, a linearized EV spatiotemporal dynamic model is developed, meeting the power requirements for large flexible EV scheduling computations. Considering the uncertainty factors inherent in EVs and transportation networks, a unified scheduling model encompassing both normal and fault states is formulated. This model relies on real-time rolling updates of information from both the transportation and distribution networks. In the normal state, the objective is to optimize economy and resilience, while the fault state aims at solely optimizing resilience. The seamless switching between these two states is realized through the introduction of state characterization parameters. The simulation results demonstrate that the proposed strategy significantly reduces computational overhead without compromising accuracy, thereby facilitating optimal resilience recovery and smooth transitions within the distribution network.