摘要
为提高模型的预测精度,引入STL算法将负荷序列分解为周期分量、趋势分量、残差分量,利用各分量训练LSTM和TCN模型。在得到LSTM和TCN模型的预测后,为进一步提高模型的预测精度,构建了集成模型对LSTM模型和TCN模型的预测值进行集成。以西班牙电力负荷数据集为例,对所提负荷预测方法进行了验证。实验结果表明,STL算法和集成模型的引入均提高了模型的预测精度,基于STL-LSTM-TCN的预测方法相较于LSTM、TCN、STL-LSTM、STL-TCN,其MAPE分别降低了2.8664%、2.1229%、0.37%、0.1%,所提负荷预测方法的预测误差最低,验证了所提预测方法的有效性与合理性。
In order to improve the prediction accuracy of the model,the STL algorithm is introduced to decompose the load sequence into periodic components,trend components,and residual components,and use each component to train the LSTM and TCN models.After obtaining the predictions of the LSTM and TCN models,in order to further improve the prediction accuracy of the models,an integrated model is constructed to integrate the predicted values of the LSTM model and the TCN model.Taking the Spanish power load data set as an example,the proposed load forecasting method is verified.The experimental results show that the introduction of STL algorithm and ensemble model both improve the prediction accuracy of the model.Compared with LSTM,TCN,STL-LSTM and STL-TCN,the MAPE of the forecasting method based on STL-LSTM-TCN proposed has decreased by 2.8664%,2.1229%,0.37%and 0.1%respectively.The proposed load forecasting method has the lowest forecast error,which verifies the validity and rationality of the proposed forecasting method.
作者
李飞宏
肖迎群
LI Feihong;XIAO Yingqun(School of Electrical Engineering,Guizhou University,Guiyang 550025,China;School of Big Data,Guizhou Institute of Technology,Guiyang 550003,China)
出处
《电子设计工程》
2023年第7期47-51,56,共6页
Electronic Design Engineering
关键词
电力负荷预测
STL分解
长短期记忆神经网络
时序卷积网络
power load forecasting
STL decomposition
Long and Short⁃Term Memory neural network
time series convolutional network