Wind power at different locations may has a significant degree of correlation. A copula function, in this paper, is employed to characterize the Joint Probability Distribution (JPD) of wind power from multiple wind farms considering their correlation. Based on this, an optimal dispatch model of power system with multiple wind farms is proposed based on Chance Constrained Programming (CCP) to describe the randomness of wind power. And a new method named Sample Average Approximation (SAA) is used to transform the chance constrians in CCP. Finally the Particle Swarm Optimization (PSO) is used to solve the dispatch model. Simulation results show the affectivity of this model and method, which will be highly useful for optimal dispatch of power system with multiple wind farms.