The historical association information mining is important for improving the accuracy and computational efficiency of short-term wind power prediction. In order to solve the problem of redundancy in the input and output variables of wind power prediction model

an index of entropy correlation coefficient(ECC)based on information entropy and mutual information is adopted. It is used to quantitatively evaluate the complex non-linear relationship between daily wind power samples of historical data and the equivalent wind speed of the next few days

and is compared with the linear correlation coefficient

rank correlation coefficient and Euclidean distance. Through intimate-samples selection

hidden layer structure optimization and network weights assignment

a modified model of short-term wind power prediction is designed to overcome the defect of the redundant degree training samples and slow convergence in traditional neural network training process

and improve the generalization ability and computational efficiency of the forecasting model. The example analysis on the measured data from a wind farm shows that the proposed method has application feasibility in improving performance of short-term wind power prediction.