Vertical water temperature simulation method based on multilayer LSTM neural network

The invention relates to a vertical water temperature simulation method based on a multilayer LSTM neural network. The method comprises the following steps of: collecting a meteorological and hydrological data set of a watershed and revising the meteorological and hydrological data set; performing standardization and normalization processing on the revised meteorological and hydrological data set to obtain a model input data set; adopting a Norton density Froude number method to carry out reservoir water temperature structure judgment to obtain a reservoir water temperature structure type; adopting an MI mutual information calculation model to input a data set and an MI value of the surface water temperature, and screening surface water temperature model input data; constructing a surface water temperature LSTM neural network prediction model, and training and verifying the model to obtain simulated surface water temperature; according to the reservoir water temperature structure type, the water temperature o