Aiming at the limited revenue sources of fast-charging electrical vehicle aggregation (FEVA), a flexible operation strategy, based on cloud-edge collaboration for fast charging stations (FCS), is proposed to participate in frequency regulation ancillary service. The proposed strategy, which takes the cloud platform and edge terminal as the core, guides the electric vehicle (EV) owners participating in frequency regulation and improves aggregator revenue while ensuring the charging experience for EV owners. Alternative charging schemes are solved for EV owners to choose from, which take the relationship between the maximum charging power of EV and the state of charge as the constraint. The cloud platform decomposes the frequency regulation signals to FCS and collaborates with them to participate in frequency regulation. The scene state machine is utilized to describe the FCS scenes and their transformation relationship. The refined mathematical models around each scene are established. While, edge terminal manages FCS locally by distributing EV power in the FCS and encouraging EV owners to end charging in advance. A deep learning model is employed to predict the next day's frequency regulation capacity for declaration. The numerical case verifies that the proposed strategy can accurately predict the frequency regulation capacity, satisfy the diversified charging demand of EV owners, and significantly improve aggregator revenue. Moreover, the cloud-edge collaborative architecture is more suitable for the frequency regulation auxiliary service.