Ultra-short-term Wind Power Prediction based on Chaos Phase Space Reconstruction and NWP

Abstract

Wind power prediction accuracy is important for assessing the security and economy when wind power is connected to the grid, and wind speed is the key factor. This article presents a future four hours prediction scheme that combined chaos phase space reconstruction with NWP method. Historical wind speed data are reconstructed as phase space vectors, which are used as the first input part of prediction model, and the NWP data at the prediction time as the second input part. Wind speed at the height of turbine hub is derived from neural network model output. To test the approach, the data from a wind farm are used for this study. The prediction results are presented and compared separately to the chaos neural network model, NWP ANN model and persistence model. The results show that the method presented in this paper has higher prediction precision.