Image Recognition Method for Secondary Circuit Terminal Based on Contrastive Learning and Improved CRNN

ChangeBridge: Patent Grants - AI & Computing (G06N)

Published March 31st, 2026

Detected April 1st, 2026

Summary

The USPTO granted Patent US12592072B1 to Sanmen Nuclear Power Co., Ltd. for an image recognition method using contrastive learning and an improved Convolutional Recurrent Neural Network (CRNN) to assess health status of secondary circuit terminals in power systems. The patent includes 11 claims and was filed on August 18, 2025, under Application No. 19303043.

View original document View source feed page

What changed

The USPTO issued Patent US12592072B1 to Sanmen Nuclear Power Co., Ltd. on March 31, 2026, covering an image recognition method for secondary circuit terminals in power systems. The method combines contrastive learning pre-training, residual neural network-enhanced feature extraction in a CRNN, and ECA-Net attention mechanism to improve recognition accuracy and detection efficiency of secondary circuit terminal block images.

This is a routine patent grant notification creating intellectual property rights for the assignee. No compliance deadlines, penalties, or required actions apply to third parties. Entities in the energy, manufacturing, or AI/ML sectors may consider reviewing this patent when developing or deploying similar image recognition technologies for power system monitoring applications.

Source document (simplified)

← USPTO Patent Grants

Image recognition method for secondary circuit terminal based on contrastive learning and improved CRNN

Grant US12592072B1 Kind: B1 Mar 31, 2026

Assignee

Sanmen Nuclear Power Co., Ltd.

Inventors

Qiuyang Lin, Congwei Wang, Luyi Zhang, Jiaqi Liu

Abstract

The present disclosure belongs to the technical field of health status assessment of secondary circuits in power systems, and specifically relates to an image recognition method for a secondary circuit terminal based on contrastive learning and an improved CRNN. The method includes: step 1: pre-training sample data of a secondary circuit terminal block of a power system through the contrastive learning; step 2: improving a feature extraction layer of a CRNN by using a residual neural network; and step 3: introducing an ECA-Net on the basis of the step 2 to construct a recognition model for the secondary circuit terminal based on the contrastive learning and the improved CRNN. In the method of the present disclosure, an image of the secondary circuit terminal block can be accurately recognized, and the accuracy of image recognition, detection accuracy and detection efficiency can be greatly improved.