In order to reduce wind power curtailments as far as possible, the expected wind power curtailments is incorporated into day-ahead dispatching formulation as a newly added minimum optimization object. The newly added optimization object is contradicted with generation costs that are intrinsic optimization object in formulation. In order to coordinate them, two optimization objects are handled respectively by fuzzy method, and a single-object optimization formulation based on maximum satisfaction extent is presented subsequently. Genetic algorithm is utilized to solve the proposed model. Large-scale integration of wind power increased uncertainties with respect to day-ahead scheduling, a constraint on risk index lost of load probability (LOLP) is proposed to replace conventional constraint on spinning reserve. Random errors on load / wind power predictions and random outages of generator units are considered in calculations of index LOLP and the expected wind power curtailments. Simulation results based on IEEE 118 test systems indicate that day-ahead dispatching formulation can provide a dispatching schedule that considers wind power curtailment and generation costs at the same time, and provide supports for system operators.