Cross-domain structure damage identification transfer learning method and device based on Grubrum angle field

The invention belongs to the technical field of structural damage identification, and particularly relates to a cross-domain structural damage identification transfer learning method and device based on a Grubrum angle field, and the method comprises the steps: collecting original time sequence signals of different structures to construct a damage identification data set, and dividing a source domain and a target domain of damage identification; carrying out dimension raising reconstruction on an original time sequence signal in the source domain damage identification data set based on the Grubrum angle field theory to convert a one-dimensional signal into a two-dimensional image signal; the two-dimensional image signals are input into a RepVGG model, the RepVGG model carries out depth feature mining on the two-dimensional image signals and carries out damage type identification, and parameters of the RepVGG model are updated; and after a small amount of damage identification data set of the target domain is