Accurate prediction of wind farms’ output is a vital part of research on issues of security and economy in situations in which a wind farm is connected to the power grid. Considering that there often exists an output correlation and complementarity between adjacent wind farms, a modeling method of wind power time series that is consistent with power output correlation and complementarity index is proposed. This article studies the historical data of three wind farms in northwest China. Based on the probability distribution characteristics, auto-correlation characteristics, and fluctuation characteristics of wind power output, a method of time series generation using Copula theory is proposed. A large number of simulation analyses of this method are carried out, the original wind farms in China and abroad are compared, and the accuracy of the method is verified. The method can well reflect the correlation of wind power output.