Large-scale wind power incorporated into power grid brings new challenges to optimal dispatch of power system. Especially, wind power at different locations may has a significant degree of correlation. A copula function was used to characterize the Joint Probability Distribution (JPD) of wind power from multiple wind farms considering their correlation. An optimal dispatch model based on Chance Constrained Programming (CCP) for power system with multiple wind farms was set up. And Sample Average Approximation (SAA) was proposed to transform the chance constrains. Finally Genetic Algorithm (GA) was employed to solve the optimal model. Simulation results indicate that copula function can well express the correlation of wind power from multiple wind farms and SAA has been a dramatic increase and improvement on the search of solution.