As an important part of a power system, the usage of wind power is increasing rapidly and playing an indispensable role in energy planning. Therefore, efforts are needed to find and improve the accuracy of wind speed forecasting and the reliability of wind energy conversion, which play a vital role in the development of wind farms. In this paper, a novel multi-objective optimization algorithm is proposed to optimize the parameters of different models, a model selection strategy is used to select the optimal hybrid models for different datasets, to improve the accuracy and stability of the forecasting model. Wind power conversion is examined based on the wind speed forecasting, and found to be a feasible method for wind farms. The numerical results show that compared with the mean absolute percentage error values of the multi-hybrid models, that of the optimal model is reduced about 3%. Moreover, the standard deviation of the absolute percentage error is decreased about 3% for wind speed forecasting. In addition, the effectiveness of the model selection is verified using the onsite wind speed data of four wind farms, and the selected model is shown to be more reliable and accurate than other models.