Intelligent detection method of rope shape abnormality of cold source interception network based on yolov5 and twin network
Wei Meng, Zhengchun Hu, Xiaolin Liu, Jun Jie Zhu, Shuai Wang, Jianwen Li · 2022 41st Chinese Control Conference (CCC) · 2022
The safety of water intakes is very important to nuclear power plants. The use of interception nets to intercept marine life is related to the safety of water intakes. This method is based on deep learning technology to detect the abnormal state of the cold source interception network. Judge whether the interception net is abnormal by comparing the historical morphology of the interception net rope, and send an abnormal warning message when an abnormality occurs to remind the management personnel to deal with and repair it in time. This method is based on independent research and development of various deep learning networks such as yolov5 and twin networks. The test results show that the recognition recall rate for abnormal changes in the shape of the interception network is more than 95%.