MEMS membrane simulation method and system based on graphic neural network

The invention belongs to an MEMS membrane structure intelligent simulation method in the field of computers, and particularly relates to an MEMS membrane simulation method and system based on a graphic neural network. The method comprises the following steps: establishing a model of boundary conditions, geometric parameters and material characteristics of an MEMS membrane to be simulated, and reconstructing a graph network model; adopting an augmented matrix algorithm to construct augmented nodes and topologies of the reconstructed graph network model to obtain graph representation; the graphic representation is input into a pre-established and trained artificial intelligence simulation model to obtain a corresponding prediction simulation result, and the prediction simulation result comprises displacement, strain, temperature, current, voltage, light intensity, magnetic flux and stress distribution of the MEMS membrane; the artificial intelligence simulation model adopts a graphic neural network. According t