Self-adaption sintering end point prediction method based on dual adversarial graph field

The invention discloses an adaptive sintering end point prediction method based on the field of double adversarial diagrams, and belongs to the field of industrial process soft measurement modeling. The method comprises the following steps: firstly, constructing a 3D adjacency matrix by utilizing a difference coefficient, and updating node feature representation by adopting a selective weighted aggregator; then, feature distribution of a source domain and a target domain is aligned through a domain discriminator, so that common features between the domains are learned; constructing a domain confusion module, confusing joint features of a source domain and a target domain by using a maximum information entropy method, forming double confrontation with a domain discriminator, and promoting the model to learn domain invariant features; and finally, predicting test data by using the trained 3D graph network feature extractor and the multi-step prediction model. According to the method, multi-step prediction of th