Abstract: Aiming at excessive smoothing in the traditional wavelet packet method to smooth wind power fluctuations,an improved wavelet packet smoothing method based on time-frequency analysis is proposed. Firstly,the set of time pairs when the maximum powerfluctuation range exceeds the fluctuation standard is found out;and the wavelet decomposition based on wavelet packet is performed on wind power to obtain its components in each frequency band. Secondly,starting from the highest frequency component,the local peakshaving and valley-filling of the components are performed only at these time pairs until the fluctuation standard is satisfied,and thus the grid-connected wind power and energy storage power command are obtained. Finally,the effect of smoothing is evaluated. Results of the case study show that this method can reduce the cumulative power and the charging-discharging switching times of energy storage power command,thus helping reduce the cost of energy storage configuration and prolong the life of energy storage while meeting the standards of wind power fluctuations.