基于TCN-LSTM和气象相似日集的电网短期负荷预测方法

为充分挖掘不同气象因素的相似日信息和输入特征蕴含的信息以提升负荷预测精度，提出一种基于时间卷积网络和长短期记忆网络组合（TCN-LSTM）和气象相似日集的电网短期负荷预测方法。首先通过Pearson系数和最大信息系数，选出与负荷强相关的气象因素；然后根据该气象因素，选取最佳相似日组成气象相似日集，以气象相似日集负荷、历史负荷、气象因素和时间因素作为预测模型的输入特征；最后，搭建TCN-LSTM预测模型，用TCN进行特征提取后，再用LSTM网络完成短期负荷预测。以中国某地区的实际历史数据进行仿真验证，结果表明所提预测方法可有效提升负荷预测精度。

Power Grid Short-term Load Forecasting Method Based on TCN-LSTM and Meteorological Similar Day Sets

In order to fully exploit the similar day information of different meteorological factors and the information contained in input features to improve the load forecasting accuracy， a short-term load forecasting method is proposed based on the combination of temporal convolution network （TCN） and long short-term memory （LSTM） network （TCN-LSTM） and meteorological similar day sets. Firstly，the meteorological factor strongly correlated with the load is selected by the Pearson coefficients and maximum information coefficients. Then， according to the meteorological factor，the best similar day is selected as the meteorological similar day sets， and load of meteorological similar day sets， historical load， meteorological factor and time factor are used as the input features of forecasting model. Finally，the TCN-LSTM forecasting model is built， TCN is used to extract the feature，LSTM is used to achieve short-term load forecasting. Simulations is made based on the real historical data of a region in China，and the results show that the proposed forecasting method can effectively improve the accuracy of short-term load forecasting.