基于GNN-LSTM融合模型的智慧公寓能耗预测与管理研究

打开文本图片集

中图分类号：TP183；TP39 文献标识码：A 文章编号：2096-4706（2025）19-0131-05

Abstract： Energy management in smart apartments is crucial for improving energy utilization eficiencyand achieving energy-saving goals.Traditional prediction methods often fail tocapture the spatial correlations between apartment units andthe nonlinear fluctuationsof energyconsumptionover time.Toaddress this，this paper proposes an innovative algorithm that combines Graph Neural Networks （GNNs） with Long Short-Term Memory （LSTM）.The proposed model uses Graph Convolutional Networks （GCNs）toeffectively extract thephysical proximityrelationships betweenapartmentunitsandleverages LSTMto capture thechanges of temporal dynamics of energyconsumption for eachunit，thereby significantly enhancing predictionaccuracy.Furthermore，this paper explores the model'sperformance indiffrentpredictiontimehorizons.Theresults indicate that the GNN-LSTMmodelmaintainsaloweror growthrate inlong-term prediction，showinggood generalization ability and practical application value.

Keywords： smart apartment; energyconsumption prediction; Graph Neural Network; Deep Temporal Model; GNN-LSTM

0 引言

建筑部门在全球能源消耗中约占 30%～40% 的比重，能耗管理已成为节能减排的关键环节。（剩余7526字）