The feedback gain in the traditional economic dispatch consensus algorithm is a fixed value, which leads to a slow convergence speed. To solve the problem, a novel distributed economic dispatch method for smart grid with time-varying feedback gain is proposed in this paper. First, by adding the direct cost, overestimation cost and underestimation cost of wind power to the objective function, an economic dispatch model considering wind power uncertainly is established, which can be flexibly scheduled under random wind speeds. Second, an improved consensus algorithm with time-varying feedback gain is proposed to solve the economic dispatch problem, which provides a design method of dynamic feedback gain instead of a fixed value and improves the convergence speed. Finally, several case studies are carried out in the IEEE 14-bus system to illustrate the effectiveness and superiority of the proposed method.