An optimal configuration method for a hybrid energy storage of a grid-connected wind storage power generation system is provided, including: performing a frequency domain decomposition on a historical wind power output, to count high-frequency and low-frequency components of the historical wind power output, and determining a rated power of the hybrid energy storage based on a probability distribution function; establishing a hybrid energy storage capacity optimization model for a full life cycle of a wind farm to minimize a net present value of an annual cost and maximize a target satisfaction rate of an output; extracting a daily typical scenario for the wind power output based on a clustering algorithm to count a time proportion of each typical scenario as an input scenario of the hybrid energy storage capacity optimization model for the full life cycle of the wind farm; and solving with a multi-objective optimization algorithm, to obtain an optimal hybrid energy storage capacity configuration scheme for the grid-connected wind storage power generation system. By optimizing a distribution of the high- and low-frequency fluctuation components between the hybrid energy storage, a battery life is extended while a fluctuation smoothing effect is effectively improved.