Currently, the widely deployment of distribution generation and electric vehicle may lead to power fluctuations of active distribution network. Along with the development of technologies in smart grid, the distribution network reconfiguration frequency may be further increased. Therefore, an hourly reconfiguration strategy is presented to minimize daily network losses and optimize power flow. Time-varying load as well as spatial-temporal EV load, wind and photovoltaic generation output are taken into consideration in the model. For the uncertain of EV load and DG output forecasting, several discrete scenarios with joint probability are divided by using the multi-scenario simulation technique. The second-order cone relaxation technique is adopted to convert the original nonconvex nonlinear problem into a mixed integer second-order cone programming problem. Finally, experimental results verify the effectiveness,robustness and accuracy of the model.