At present, robust dispatch is widely used to solve the economic dispatch problem of power system with wind power. Robust dispatch can ensure that constraint conditions are satisfied for all possible values of the wind power output, but it is too conservative and greatly reduces the optimality of objective function. Based on the traditional robust dispatch model of power system, this paper firstly analyzes the influence of wind uncertainty on power system dispatch. Then, taking coal consumption cost, risk cost and power flow violation as optimal targets, a multi-objective optimization model with power dispatch conservativeness (the confidence level of wind power output) as decision variable is proposed. Finally, the model is solved by a two-layer optimization algorithm. In the case study, the proposed method, the traditional spinning reserve method and the traditional robust dispatch method are compared. It is found that the conflict between economy and robustness of robust dispatch can be coordinated by optimizing the conservativeness. Though the proposed method takes the longest time, it can obtains the optimal comprehensive satisfactory degree.