NMOSFET device performance prediction method based on BP neural network

The invention relates to the technical field of semiconductors and the field of artificial intelligence, and discloses an NMOSFET device performance prediction method based on a BP neural network. The size of the device is determined according to actual needs, TCAD simulation software is used for simulating the device, and the performance of the device under different conditions is obtained by changing structural parameters and external pressure applying conditions of the device. And substituting the structural parameters and the external conditions as independent variables and the performance of the NMOSFET as dependent variables into a BP neural network prediction model for continuous training, and comparing model evaluation indexes to obtain a model with the highest prediction precision. The model can be used for predicting the performance parameters of the NMOSFET, and the high-precision prediction model can also improve the efficiency of determining the simulation performance of the device; meanwhile, a