摘要
为了解决钢丝绳漏磁检测过程中抖动信号的干扰问题并实现缺陷信号的准确分类,提出一种基于改进Swin Transformer的多通道漏磁信号分类方法。通过环形阵列传感器采集多通道漏磁信号,并结合数据增强技术,构建一个包含4类信号的漏磁信号数据集,用于模型训练与测试。试验结果表明,所提出的模型在各项评价指标上均优于传统的一维卷积神经网络(1D-CNN)、长短时记忆网络(LSTM)和门控循环单元(GRU),具有强大的特征提取与分类能力。最后,利用可视化工具对模型决策过程进行了解释。
To address the interference from jitter signals during magnetic flux leakage detection of steel wire ropes and achieve accurate classification of defect signals,this paper proposes a multi-channel time series signal classification method based on an improved Swin Transformer.Multi-channel magnetic flux leakage signals are collected using an annular array sensor,and in conjunction with data augmentation techniques,a dataset comprising four types of magnetic flux leakage signals is constructed for model training and testing.Experimental results demonstrate that the proposed model outperforms traditional one-dimensional convolutional neural networks(1D-CNN),long short-term memory networks(LSTM),and gated recurrent units(GRU)across various evaluation metrics,highlighting its superior feature extraction and classification capabilities.Finally,the decision-making process of the model is elucidated through the use of visualization tools.
作者
姜帅
胡而已
孙燕华
JIANG Shuai;HU Eryi;SUN Yanhua(School of Mechanical Science and Engineering,Huazhong University of Science and Technology,Wuhan 430074,China;Information Research Institute of the Ministry of Emergency Managment,Beijing 100029,China)
出处
《无损检测》
2026年第2期13-20,共8页
Nondestructive Testing
基金
国家自然科学基金(52275532)
交通运输部纵向项目(SXHXGZ-2021-2)
国家重点研发纵向项目(2021YFF0501000)。
