Abstract: A new short-term wind speed prediction method by using particle swarm optimization extreme learning machine based on principal component attribute reduction clustering is proposed. Considering the influence of different attribute characteristics on the change of wind speed,the principal component analysis method is used to calculate the eigenvalues of each component,and the components with high variance contribution rate are selected. Then the wind speed samples are grouped by k-means clustering method,and then the extreme learning machine is optimized by particle swarm optimization algorithm. Further,the wind speed combination prediction model is constructed. Finally,the experimental prediction and comparisons are carried out with the measured historical data of wind farm. The results show that the method has high prediction accuracy.