Water consumption prediction method based on GNN and LSTM combined model

The water consumption prediction method based on the GNN and LSTM combined model comprises industrial park enterprise water consumption prediction of the GNN (graph neural network) and LSTM (long and short term memory network) combined model, and comprises the following steps: S1, obtaining historical water consumption and related influence factor data of enterprises in a park, including time identifiers and meteorological identifiers, and carrying out data preprocessing; s2, a combination model based on GNN and LSTM is constructed, in the model, the LSTM model is used for extracting enterprise water consumption time features, the GNN model is used for extracting enterprise relation features, the GNN-LSTM superposition model is used for extracting combination features, and after feature splicing, the features serve as input to obtain a water consumption predicted value through a full connection layer; s3, performing model training by using historical data, updating model parameters, and obtaining a combined p