Two-phase heat transfer prediction method based on physical information neural network

The invention relates to the technical field of fluid dynamics, and discloses a two-phase flow heat transfer prediction method based on a physical information neural network, which comprises the following steps of: dividing a computational domain into a plurality of units, obtaining a sample point coordinate by adopting a Legendre-Gauss-Labotto sampling method, inputting the sample point coordinate into an initialized full-connection neural network, and obtaining a two-phase flow heat transfer prediction result; obtaining data of speed, pressure, temperature and phase variables, calculating partial derivatives and unit matrixes of the data to space coordinates and time, and constructing a control equation set based on the data of the speed, the pressure, the temperature and the phase variables; embedding the control equation set into the full-connection neural network, and obtaining a total loss function based on the control equation set, the initial condition, the boundary condition and the boundary derivati