Abstract: As to such features of wind power as counter peak-shaving and prediction deviation, a dynamic optimal dispatch model of power system combined with wind power, hydro power and thermal power is proposed, which can be optimized by hierarchical scheme. It is solved by layered optimization such as thermal power start-up mode optimization layer which can avoid redundancy and low-load operation of thermal power, hydropower peak-shaving optimization layer which can undertakes load as far as possible while restraining the load and the fluctuation of wind farm, and wind-thermal power optimization layer which takes into account the additional cost from prediction deviation of wind power at different period and makes the total cost of system minimum. And tracking center locus interior point method is used to solve the second and third optimization layers. The proposed models are validated by the simulation results based on a power grid containing a wind farm and several hydropower stations. Simulation results show that proposed optimized model and method offer an effective approach to solve dynamic optimal dispatch of power system combed with wind power, hydro power and thermal.