Wind power forecast is of great significance for the safe and stable operation of wind farms and power grid dispatching. At present, the selection of wind power short-term forecast indicators is unreasonable and the forecast accuracy is low. Aiming at these problems, a short-term wind power forecast model based on Pearson correlation coefficient (PCC) and radial basis function (RBF) neural network is proposed. Firstly, three indicators closely related to wind power, e.g., current, temperature and wind speed, are selected by PCC. Then, these three indicators are used as the input of the forecast model for RBF samples training and short-term forecast of wind power. The results show that the proposed forecast model has smaller forecast error and higher prediction accuracy. It can meet the requirements of short-term wind power prediction and has a wide application prospect.