【摘要】 考虑风力发电具有随机性和不稳定性，为准确预测风速，提出一种基于完备集合经验模态分解和双向门控单元网络相结合的短期风速组合预测方法。首先，采用完备集合经验模态分解，将原始风速序列分解为若干个具有较强规律性的子序列，以减少不同特征尺度序列间的相互影响；然后，利用样本熵来评估风速子序列的复杂度，将复杂度相近的子序列组合为一个新序列，以减少输入到神经网络的模型数量；最后，将新组合的子序列分别输入到双向门控单元网络中进行预测，得到各子序列的预测结果，叠加得最终的风速预测结果。实例预测结果表明，所提出的风速预测方法具有较高的精度和运行效率。更多还原

【Abstract】 Considering the randomness and instability of wind power generation, to improve the accuracy of wind speed forecasting, a short-term wind speed combined forecasting method combining complete ensemble empirical mode decomposition(CEEMDAN) and bidirectional gated unit network(BiGRU) is proposed. Firstly, the CEEMDAN is used to decompose the original wind speed sequence into several sub-sequences with strong regularity to reduce the mutual influence between different feature scale sequences; then the sample entropy(SE) is used to evaluate the complexity of the wind speed sub-sequences. The subsequences with similar degrees are combined into a new sequence to reduce the number of models input to the neural network. Finally, the newly combined subsequences are input into the BiGRU for prediction, and the prediction of each subsequence is obtained. The example prediction results show that the proposed wind speed prediction method has high accuracy and operating efficiency.更多还原