Abstract: This paper proposes a short-term wind speed prediction model based on fast correlation reduction based on convolutional memory network. Firstly,the information entropy of the correlation degree of each wind speed series and its attribute series is calculated,and the fast correlation filtering algorithm is used for attribute reduction to reduce the attribute dimension and delete redundant attributes. For wind speed attribute matrix samples,Squeeze-and-Excitation Networks are used to construct attribute representation sequences. The sequence spacing is used as sample similarity to realize sample set optimization and reconstruction by Affinity Propagation Clustering. Secondly,a convolutional memory network is built and used to mine deep features and short-term predictions. Finally,by predicting the wind speed of the actual wind field and comparing the measured data,the accuracy and effectiveness of the prediction model are verified. The results show that the method proposed in this paper has a great advantage in the optimization of wind speed attribute data. By retaining closely related attribute information,the prediction accuracy is improved.