Abstract: Wind power farms（WPFs） have lower accuracy of power prediction and generation schedule curves. The electrochemical energy storage system（EESS） can help the WPF to improve the schedule curve tracking ability, but its unit investment per unit is expensive. It is necessary to explore a capacity planning method of EESS, which can consider deviation correction objective of generation curves in WPFs and the capacity miniaturization demand of EESS. A capacity planning method for the small-scale EESS is proposed, which considers the return-to-zero characteristics of short-time net deviation power of the unscheduled deviation power curve in WPFs and faces the reduction of deviation assessment power. The excessive deviation power amplitude and return-to-zero characteristics of net deviation power that affect the generation curve deviation correction effect（GDCE） are defined. A relationship model between GDCE, two characteristics and energy storage capacity is established, as well as a capacity planning model of EESS that comprehensively considers the technical constraints of the excessive power ratio and the economic objective of rate of return on investment. The calculation example shows that there is a strong correlation between GDCE and characteristics, and the proposed method can consider the characteristics, meet the requirements of deviation correction for the power generation curve, and can provide technical support for EESS with moderate investment in WPFs.