Disordered charging of clustered Electric Vehicles (EV) leads to increased network loss and voltage drop in distribution network, which seriously affects the safe and reliable operation of power grid. Aiming at the problem that the single charging demand only considering slow or fast charging cannot effectively describe the charging characteristics, which leads to insufficient charging scheduling efficiency, this paper fully considers the charging demand of different modes in multi regions. Based on the National Household Travel Survey (NHTS) data, the key parameters of residential slow charging are deeply analyzed. Based on the measured data of cleaned public fast charging stations, the charging behavior characteristics of public fast charging are depicted, and the EV charging load models in residential slow charging and public fast charging are constructed. To solve the problem that the EV charging scheduling model is a mixed integer nonlinear programming model with non-convexity and nonlinearity, the second-order cone programming theory is introduced to relax the original model into a convex programming model. The optimal scheduling strategy for EV charging is established with the minimum network loss as the optimization goal. The results show that compared with disordered charging, the proposed strategy reduces the network loss by 6.74%, and increases the minimum node voltage from 0.9500 to 0.9671, so that the node voltage is in the fluctuation range [0.95–1.05], which ensures the safe and reliable operation of the grid.