Abstract: Since the current research on charging and discharging service mode only considers EV load guidance from the perspective of time-shifting, a “reservation/on-demand” mode is established incorporating the consideration of the regularity of users’ grid-access behaviors. In this mode, electric vehicle aggregator（EVA） can predict the charging needs of users with regular grid-access behaviors through service reservations, thereby improving the guidance effect of EVAs. Firstly, the business process of “reservation/on-demand” service is elaborated. Secondly, considering the comprehensive economic effect and convenience effect, with the goal of maximizing user’s utility, the EVAs can construct optimal service purchase strategies for various users and analyze purchase intentions. Finally, from the perspective of EVA, this paper constructs a utility-based user service selection behavior analysis model to help EVA grasp the influence of different user choices, and uses the Stackelberg game method to process EVA price guidance strategy. Analysis shows that, compared with the existing simplistic service model, the “reservation/on-demand” service mode can not only improve service quality, but also further reduce the power purchase cost of EVA.