Abstract: The integration of massive wind power brings significant technical challenges to power system operation. The randomness and volatility of wind power output increase the risk of insufficient transmission or regulation capacity, and the traditional determined economic dispatch methods are no longer applicable in this condition. First, a chance constrained stochastic economic dispatch model is proposed based on Gaussian mixture distribution. In this model, the minimum reserve constraints and transmission capacity constraints are represented by chance constraints, and the affine control strategy of regulation units is adopted to balance the real-time wind power fluctuation. And then, a convex relaxation based iterative solution method for chanceconstrained programming model is proposed, by which, stochastic dispatch model is converted into a quadratic programming model to achieve the fast and effective solution. Taking data of 25 wind farms in Southwest China and the IEEE 118-bus test system as the analysis cases, the fitting accuracy of Gaussian mixture distribution for wind power forecasting error and the efficiency of the proposed method are verified.