For breaking the bottleneck that further improvement of the wind power prediction accuracy encounters, an idea of making use of the intensive control of the Smark Park to depress wind power prediction error is proposed. Firstly, a detailed Smark Park simulation model is established and the port power characteristics of charge/discharge unit are analyzed. The relationship between the battery storage energy and the state of charge (SOC) is obtained. Based on a real wind power output and its prediction data, this paper analyzes the relationship between the target of depressing wind power prediction error and the factors such as the size of Smart Park, initial state of charged distribution, wind power installed capacity and wind power forecasting technique level. The results show that the coordinated operation of wind farms and Smart Park can depress wind power prediction error to a great extent, and it is an economic and feasible way to turn the undispatchable wind farm into dispatchable energy sources.