【摘要】 由于邻近多风电机组间存在复杂的关联关系，深度挖掘多风电机组的空间特征有利于提高风电功率的预测精度。因此，提出了一种考虑多风电机组关联特性的超短期功率预测方法。首先，基于能量谷优化算法对变分模态分解的关键参数进行优化，将原始风电功率数据分解为多个利于预测的模态分量；随后，在双向门控循环单元时序预测网络中引入时序注意力机制，充分提取多风电机组间的复杂联系，从时空角度对各模态分量进行精准预测；最后，对各模态分量预测值进行重构得到多台风机的风电功率预测。实验结果表明，与其他预测模型相比，所提方法不仅能有效提高多风电机组的超短期功率预测精度，同时也能缩短训练时间。更多还原

【Abstract】 Due to the complex correlations among neighboring wind turbines, deeply exploring the spatial features of multiple wind turbines is beneficial for improving wind power prediction accuracy. Therefore, an ultra-short-term power prediction method considering the correlation characteristics of multiple wind turbines is proposed. Firstly, the energy valley optimization(EVO) algorithm is employed to optimize the key parameters of variational mode decomposition(VMD), which decomposes the original wind power data into multiple mode components that are more conducive to prediction. And then, a temporal attention mechanism is introduced into the bidirectional gated recurrent unit(BiGRU) sequence prediction network to fully extract the complex relationships among multiple wind turbines and accurately predict each mode component from a spatiotemporal perspective. Finally, the predicted values of each mode component are reconstructed to obtain the wind power prediction for multiple turbines. Experimental results show that, compared to other prediction models, the proposed method not only effectively can improve the ultra-short-term power prediction accuracy for multiple wind turbines, but also can reduce the training time.更多还原