A frequency regulation method for a large number of electric vehicles (EVs) in an isolated grid with high permeability renewable energy sources is proposed. First, a disturbance observer is designed for the system's order reduction model to generate additional frequency control signals for clustered EVs. This order reduction model is obtained by combining the changes in load, wind power, photovoltaic system and clustered EVs, thus generating a lumped disturbance estimated by the disturbance observer. Second, a robust model predictive control method based on the Tube model is proposed to provide effective control signals to improve the responsiveness of clustered EVs. The control signals are generated to obtain the minimum frequency deviation error by means of the minimum control actions while considering various physical constraints on the system operation. Third, the influence of time delay on communication link is studied through the stability analysis, and the time delay margin is obtained. Finally, through simulation analysis, the effectiveness of the proposed method is verified, and the advantages of the proposed method over traditional model predictive control, fuzzy proportional integral control and linear quadratic regulator control are also verified.