Energy consumption prediction of cold source system based on GraphSAGE
Zhiwen Chen, Qiao Deng, Zhengrun Zhao, Bei Sun, Tao Peng, Chunhua Yang
Abstract
The energy-saving optimization control of central air-conditioning is a necessary way to achieve energy saving in buildings, and the energy consumption prediction of the cold source system is the prerequisite for achieving energy-saving optimization control. Due to the strong coupling between the central air-conditioning systems and the complicated time relationship between the operating data, this paper uses the graph neural network method to predict the energy consumption of the cold source system. First, KNN is used to find the association information between the operating data, and an association graph is constructed. Then, this paper uses the graph neural network GraphSAGE to predict the energy consumption of the cold source system. The performance of this method is significantly better than that of 1D-CNN and RF models when there are fewer training samples.