Wind power prediction is an effective means to reduce the phenomenon of abandonment of wind power. This paper provides a means that introduce turbulence value IT in Fuzzy C-Means algorithm (FCM), which can further enhance the similarity between the predicted samples and the training samples, avoid the negative impact of wind power fluctuation due to the decrease of the training samples. This method is verified on MATLAB platform, where the prediction result is generated based on measured dates from a wind farm in Shanxi by Support Vector Machine (SVM) and FCM. From the consequence we can see that FCM-IT-SVM can be able to strengthen the similarity of wind power, and reduce the error of wind power prediction effectively.