The generation power of wind farms is very uncertain and difficult to correlate. It affects the distribution of uncertain power flow in a power system. As a tool for calculating uncertain power flow, Interval Power Flow (IPF) should take into account the uncertainty and correlation of wind power, so as to accurately obtain the interval distribution characteristics of the unknowns in power flow. This paper uses the correlation angle of a joint sampling area to quantify the interval correlation of wind power, constructs an interval power flow model considering the interval correlation, and proposes an Optimal Scenario Algorithm based on Affine Transformation (OSA-AT) to solve it. First, the affine transformation is used to transform the correlated wind power output interval distribution into independent interval variables. Secondly, the optimal scenario method is used to transform the interval power flow into a series of nonlinear optimization problems. Finally, the interior point method is used to calculate the maximum and minimum value of the unknowns in power flow, known as the interval distribution. The numerical results of IEEE 14- and 118-bus systems indicate that the proposed method can deal with the correlation of interval variables accurately. Compared with Monte Carlo (MC) method, the computing efficiency can be improved by several dozen times. This work is supported by Key Science and Technology Program of China Southern Power Grid Company (No. GXKJXM20170522).