Specific Emitter Identification (SEI) technology enhances communication security in the industrial Internet of Things sector. However, challenges in obtaining high-quality labeled data for wireless devices restrict the application of SEI technologies. This has prompted the investigation of SEI methods that leverage unlabeled data. To this end, this paper proposes an unsupervised domain adaptation method known as the Residual Independence Criterion (ResIC). Specifically, we use the Hilbert-Schmidt independence criterion to assess the similarity of samples across domains, combining it with a residual network to enhance feature extraction with inter-class separation. This approach enhances the transfer of knowledge from the source to the target domain, significantly improving model performance in target domains, which contains a large volume of unlabeled data. In four transfer scenarios using WiFi datasets constructed in a laboratory environment, this paper compares the proposed method with a baseline method and three unsupervised domain adaptation methods. The results demonstrate that ResIC reduces the target domain’s reliance on high-quality labeled data.