With the access of large-scale renewable energy to the microgrid, its uncertainty directly affects the optimal scheduling of the microgrid. In this paper, aiming at maximizing the generation profits of microgrid, an optimization model of day-ahead generation schedule for the microgrid is constructed, with consideration of the energy storage, demand response and error handling of renewable generation prediction. Then an improved artificial bee colony algorithm based on crossover and mutation operations is proposed to solve this problem. In the employed bee phase and onlooker bee phase, crossover and mutation processing is introduced to improve the neighborhood search strategy, in order to ensure the diversity of offspring population. In the scout bee phase, an initialization mechanism based on global search is constructed to improve the ability of searching global optimal solution. Finally, the simulation results demonstrate the effectiveness of the model and the superiority of the algorithm.