Reservoir level prediction method based on graph neural network

The invention provides a reservoir level prediction method based on a graph neural network, and provides a reservoir level prediction model combining LSTM and GCN. Potential characteristics of long-term and short-term time sequence dependence containing multivariable sequences are extracted through LSTM, on the basis, a graph structure learning module is introduced, and each variable is regarded as a node to learn the interaction relationship between different variable characteristics. And finally, fusing the topological information of the graph structure into the potential features through GCN, and carrying out final water level prediction. 本发明提出一种基于图神经网络的水库水位预测方法，提出了一种联合LSTM和GCN的水库水位预测模型。通过LSTM提取包含多变量序列的长期、短期时序依赖的潜在特征，在此基础上，引入图结构学习模块，将每个变量视作一个节点来学习不同变量特征之间的相互作用关系。最后，通过GCN将图结构的拓扑信息融入潜在特征中，并进行最终的水位预测。