AUTHOR=Zhao Ning , Fang Yongyi , Wang Siying , Li Qian , Wang Xiaonan , Feng Chi 

TITLE=Research on the identification method of cable insulation defects based on Markov transition fields and transformer networks

JOURNAL=Frontiers in Physics

VOLUME=Volume 12 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/physics/articles/10.3389/fphy.2024.1432783

DOI=10.3389/fphy.2024.1432783

ISSN=2296-424X

ABSTRACT=To enhance the accuracy of cable insulation defect identification and the robustness of the algorithm against noise, this paper proposes a method for identifying cable insulation defects based on the Markov transition field (MTF) and the Transformer network. Firstly, the algorithm performs modal transformation on the time series data acquired by the ultrasonic probe through MTF, generating corresponding images. Secondly, these image data are fed into a pre-trained Transformer network to achieve automated feature extraction. Then, a multi-head attention mechanism is introduced, which can assign weights to the features extracted by the Transformer network, thereby emphasizing the most critical information for the identification task Ultimately, more accurate defect identification is achieved based on the weighted features. Compared with traditional image processing and recognition methods, the method proposed in this paper demonstrates higher accuracy and stronger noise resistance in the task of cable insulation defect identification.