基于Informer的城市住宅区负荷预测研究

Research on Power Load Forecasting for Urban Residential Areas Based on Informer

对城市住宅区的电力负荷进行长时间序列的预测能够为电力资源提前分配提供最优的技术支持,从而确保电力供需之间的平衡.为了提高对于长序列负荷预测的能力,将Informer模型应用于城市住宅区负荷预测领域;为了提升城市住宅区负荷预测的精度,先采用交互信息从原始数据中筛选出与住宅区历史负荷相关性较强的特征数据并进行数据增强,然后使用Log-Cosh损失函数对Informer模型进行训练.测试结果表明Informer模型比长短时记忆网络的预测精度更高,证实了所提Informer方法的可行性.

The long-term sequence forecasting of the power loads in urban residential areas can enable the optimal distribution of electric power resources in advance and ensure the balance between power supply and demand. In order to improve the ability of long-sequence load forecasting, the Informer model is applied to forecasting the power loads of urban residential areas. The Informer model employs its unique self-attention mechanism and encoder-decoder architecture to accurately capture the long-term correlation between the input and output of the load sequence. In order to further improve the accuracy of load forecasting for urban residential areas, characteristic data that are strongly related to the historical loads of residential areas is screened out using mutual information,and then the data is enhanced,and then the Log-Cosh loss function is used for the Informer model conduct training. The test results on the data set show that the prediction accuracy of the Informer model is higher,which confirms the feasibility of the proposed method.