The number of electric vehicles is mounting in the process of achieving "dual carbon" target. And their charging and discharging behaviours significantly impact the operation of power grid. In response to the numerous control instructions given to schedule every single electric vehicle(EV), considering from EV side and distribution network side, a dispatchable capacity assessment index and a benefit assessment index are proposed. And a comprehensive dispatchable potential assessment model for EVs is established. Then, EVs can be divided into different groups and those in dispatch priority zones will be given schedule instructions preferentially. To alleviate the difficulty brought by EV mobility between regions to the model solving, an EV grouping method based on W-GAN and k-means clustering algorithm is used. Based on three peak load regulation cases, the dispatchable potential and dispatch performance of four EV groups are analysed. The simulation results show that the dispatchable capacity assessment index and the dispatching benefit assessment index of the proposed method are reasonable. Giving scheduling priority to the divided EV dispatch priority zones selected by the integrated dispatchable potential assessment model can accomplish peak load regulation with less dispatching times.