According to the principle that reactive power should be compensated locally， the reactive power needed by a wind farm should be provided by the wind farm reactive compensation devices as much as possible. By considering the valve-point effect of conventional generators and the discreteness of asynchronous wind turbine reactive compensation devices， the dynamic optimal power flow（DOPF） of the wind power integrated system is a typical complex multi-constrained mixed-integer non-convex and non-linear programming problem. The coupling between different stages is taken into account and the minimum cost of all conventional units in the whole stage is set as the objective function to establish a model for DOPF of the wind power integrated system. A novel shuffled frog leaping（SFL） algorithm is used to solve the DOPF problem and the detailed realization of this algorithm is given. Simulation results of the improved IEEE 30-bus system with the SFL algorithm are better than those of the particle swarm optimization（PSO） algorithm， which shows the effectiveness of the method proposed.