In order to study the impact of the uncertainty of wind power forecast errors on the economic dispatch of the power system, firstly, the improved Gaussian mixture model is used to characterize the uncertainty of wind power forecast errors. According to the model, using Latin hypercube sampling generate the scene, and the improved K-means clustering is used to reduce the scene to obtain the initial scene and probability. Since the initial probability has a certain degree of randomness, the comprehensive norm is used to obtain its uncertain set. Then, to promote the consumption of wind power, the impact of energy storage and time-of-use electricity prices are considered at the same time during the dispatch process, and finally a min-max-min two-stage economic dispatch model is constructed. The first stage takes the start-up and shutdown cost of the unit as the goal, and the second stage searches for the optimal operating cost under the worst probability distribution based on the uncertainty set, and uses the column constraint generation algorithm to solve the model. Finally, the IEEE 39-node system is used to prove the effectiveness of the model and algorithm.