基于VMD-PCA和TCN模型的短期电力负荷预测

Short-term power load forecasting based on variational mode decomposition-principal component analysis and temporal convolutional network

为提高电力负荷预测的准确性以降低后期电力备用储能建设的成本,需采取合理精确的预测模型预测未来负荷数据,文中提出一种基于变分模态分解(VMD)结合主成分分析(PCA)与时间卷积网络(TCN)组成的电力负荷预测模型。首先,为了提高抗噪性和分解效率,采用VMD对原始负荷序列进行分解,分解所得的模态分量通过计算样本熵值(SE)进行复杂度的近似分类,对新序列组分别建立预测模型;然后,采用主成分分析法做特征提取,提取出对预测目标影响较大的影响因素作为模型的输入向量。时间卷积网络作为深度卷积网络的改进算法,在预测精度和时间上都具有较大的优势,在深度学习领域得到了很多的关注,采用该模型进行短期电力负荷预测,最终结果同其他模型的结果相比误差最小,证明了该预测模型的精确可靠性。

It is necessary to adopt a reasonable and accurate prediction model to predict future load data,so as to improve the accuracy of power load prediction and reduce the cost of later power reserve energy storage construction.In this paper,a power load prediction model based on variational mode decomposition(VMD)-principal component analysis(PCA)and temporal convolutional network(TCN)is proposed.In order to improve the noise resistance and decomposition efficiency of the model,the original load sequence is decomposed by VMD.The decomposed modal components are subjected to approximate complexity classification by calculating the sample entropy(SE),and the prediction models are established for the new sequence groups.And then,the PCA method is used to extract the features,and the factors which have great influence on the prediction target are extracted as the input vector of the model.As an improved algorithm of deep convolution network,TCN has great advantages in prediction accuracy and time,and has attracted much attention in the field of deep learning.When this model is applied to short-term power load forecasting,the error of its final forecasting result is the smallest in comparison with that of other models,which proves the accuracy and reliability of the forecasting model.