基于混合分解技术的鲁棒极限学习机的风速预测

提出一种基于混合分解技术的改进鲁棒极限学习机的风速预测模型。混合分解技术的特殊性在于采用变分模态分解,把互补集合经验模式分解所产生的高频固有模态函数进一步分解为多个模态分量,以提高短期风速预测的精度。然后对混合分解技术分解得到的全部风速分量分别建立鲁棒极限学习机模型进行预测,并采用一种改进鲸鱼优化算法对鲁棒极限学习机的参数进行微调。最后,根据西班牙某-风电场实际风速数据进行风速多步短期预测。实验结果表明:基于混合分解技术和改进鲸鱼优化算法优化鲁棒极限学习机的组合预测模型在风速预测.中取得较好的预测效果。

Wind speed prediction of improved robust extreme learning machinebased on hybrid decomposition technique

A kind of wind speed prediction model of improved outlier robust extreme learning machine( ORELM) based on hybrid decomposition technique is proposed. The speciality of hybrid decomposi-tion technique is that variational mode decomposition( VMD) is used to further decompose the high -frequency intrinsic mode functions ( IMFs ) generated by complementary ensemble empirical modedecomposition( CEEMD) into a number of modes in order to improve the forecast accuracy. The windspeed components obtained by the hybrid decomposition technique are separately established ORELMmodel to forecast, and an improved whale optimization algorithm( IWOA) is used to make trimming ofthe parameters of the ORELM. Finally, wind speed multi-step short-term prediction is carried outaccording to the actual wind speed data of a certain Spanish wind farm. The experimental results showthat the combined forecasting model based on the hybrid decomposition technique and ORELM trainedby IWOA has higher prediction accuracy than other forecasting model.