Small space model real-time training method based on semiconductor device structure

The invention discloses a small space model real-time training method based on a semiconductor device structure, and the method comprises the steps: S1, determining a target semiconductor device structure at a user interface, inputting the target semiconductor device structure into a cloud database, and obtaining a numerical value X of each parameter of the target semiconductor device structure; s2, on the basis of similarity, setting an adaptive threshold value, and automatically screening a small-space training sample data set in a threshold value space in a cloud database; s3, transmitting the small space sample data set to an edge end computing device, and constructing a prediction model from the structure to the electrical performance in real time on the edge computing device by using a machine learning algorithm; and S4, inputting the structure X stored at the edge end into the model in the step 3, and outputting the electrical performance of the device. According to the invention, a small space is sele