For the spread of wind power generations, the technology of reducing fluctuations in wind power output is indispensable. In this research, we perform computational analyses of a wind power generation system with storage battery, and propose a novel approach for deriving the proper battery capacity. This research is to achieve sufficient smoothing of wind power output and efficient use of wind power energy. In the proposed method, numerically investigating the trade-off relationship among the smoothing effect, battery capacity, and output efficiency characteristics, we can derive the battery capacity on the basis of cost/performance. Furthermore, we performed the investigation of the system using wind power prediction method, i.e., Just-In-Time Modeling. By predicting the wind power output, we can properly update the smoothing time constant according to the short-term wind condition, which results in the reduction of battery capacity. To verify the proposed approach, we also present some computational results based on the measured data of small wind power generation system.