Pattern-Based Wind Speed Prediction Based on Generalized Principal Component Analysis

Short-term wind speed prediction plays an important role in large-scale wind power penetration. However, there is still a large gap between the requirement of prediction performance and current techniques. In this paper, we propose a pattern-based approach to short-term wind speed prediction. It is well accepted that wind varies in different patterns in different weather conditions. Thus, we should use different models to describe these patterns, whereas most current works conduct wind speed prediction with a single model. Based on this observation, we introduce generalized principal component analysis to automatically discover the patterns hidden in the historical data of wind speed. Then we train a predicting function for each pattern and combine their outputs for the final prediction. Experimental results show that the proposed approach performs better than the clustering-based approach, a single model, and persistence forecasting.