The randomness of wind farm stroke output will cause the power fluctuation of grid-connected

and the traditional energy storage method will produce the situation of over-charge and over-discharge of energy storage when grid-connected. This paper proposes a layered optimization strategy of hybrid energy storage under charge state considering the prediction error of wind power. Firstly

the wind power data is analyzed

and the advanced control strategy is proposed by using the predicted power information. The dynamic interval of the prediction information is determined through the probability and statistics distribution of the charge/discharge switch

and the total scope of the mixed energy storage is solved. Then

the wind output was decomposed into frequency

and the total scope was decomposed into supercapacitor scope and storage battery scope by the adaptive noise ensemble empirical mode decomposition(improved complete ensemble empirical mode decomposition with adaptive noise

ICEEMDAN). Then

the charge/discharge capacity of energy storage and the avoidance of over-charge and overdischarge of energy storage were considered

and the optimization was carried out by the double fuzzy controller.Finally

the energy storage is compensated according to different working modes through the energy storage action assist collaboration to optimize the charged state of the energy storage. A wind farm in Xinjiang of China was taken as an example to verify the effectiveness of the proposed strategy.