Abstract: In this paper, based on the variational mode decomposition (VMD), outlier-robust extreme learning machine (ORELM) and error correction (EC), a combined wind speed prediction model (VMD-ORELM-EC) is proposed to improve the accuracy of ultra-short-term wind speed prediction. Firstly, the original wind speed series are decomposed by the VMD, and the obtained decomposition sub-series are used to build the ORELM sub-models. The prediction results of each sub-model are calculated to obtain the preliminary prediction series. Then, by subtracting the preliminary prediction series from the original wind speed series, the error series of the model can be determined. Accordingly, by employing the VMD and the ORELM, the error prediction series can be obtained. Finally, the preliminary prediction series are combined with the error prediction series to determine the final wind speed prediction series. The proposed VMD-ORELM-EC model is further employed to analyze the field-measured wind speed data obtained from the Beijing anemometer tower. The results show that the model can effectively exploit the characteristics of wind speed series and has high prediction performance in ultra-short-term wind speed prediction.
Keywords: ultra-short-term wind speed prediction; variational mode decomposition (VMD); outlier-robust extreme learning machine (ORELM); error correction (EC)
合肥工业大学学报自科版