【摘要】 负荷预测对于电力企业制定未来调度计划十分重要。为了进一步提高预测精度，充分挖掘负荷数据中时序特征的联系，提出一种卷积神经网络(Convolutional Neural Networks, CNN)、门控循环单元(Gate Recurrent Unit, GRU)和多元线性回归(Multiple Linear Regression, MLR)混合的多频组合电力负荷预测模型。该模型先对时间序列的负荷数据进行集合经验模态分解(Ensemble Empirical Mode Decomposition, EEMD),并将其重构为高低两种频率；同时在高频中引入影响因子较大的气象因素，使用CNN-GRU模型预测，低频部分使用多元线性回归进行预测；最后将各个模型得出的预测结果叠加，得到最终预测结果。仿真结果表明，相对于其它网络模型，提出的混合模型具有更高的预测精度，是一种有效的短期负荷预测方法。更多还原

【Abstract】 The load forecasting for the future scheduling plan ofelectric power enterprises is very important. In order to further improve the prediction accuracy and excavate the relationship of time series features in load data, this paper proposes a multi-frequency combination power load forecasting model based on convolutional neural network(CNN),gate recurrent unit(GRU) and multiple linear regression(MLR). In this model, the load data of time series were first decomposed by ensemble empirical mode decomposition(EEMD),and then reconstructed into high and low frequencies. At the same time, for the high frequency part, the GRU-CNN model was used for prediction, while for the low frequency part, MLR was used. Finally, the prediction results obtained from each model were superimposed to form the final forecast results. The simulation results show that the proposed model has higher forecast accuracy compared with other network models and it is an effective method for short-termloadforecasting.更多还原