华中科技大学学报（自然科学版）2025，Vol.53Issue(7)：16-23,8.DOI:10.13245/j.hust.250268
基于VGAF与混合时序网络的电缆故障诊断方法
Cable fault diagnosis method based on vgaf and hybrid timing network
摘要
Abstract
A cable fault diagnosis method based on the variational modal decomposition Gram's angle field(VGAF)and hybrid convolutional time series network algorithm(CLT)was constructed.Firstly,the historical fault data in the input Gram's Angle field(GAF)was optimised by variational modal decomposition(VMD),and the spurious components were eliminated according to the centre frequency and the threshold β.Then,the optimised data was converted into a two-dimensional feature map after the normalisation process using the GAF,and the multi-feature information was selected and inputted into the CLT network.Finally,the predicted value was outputted in the hybrid time-series network,and the evaluation metrics λ1 and λ2 were used in the model to evaluate the diagnostic accuracy of the network.Verified by the example data,the proposed combination model in this paper can reduce mean absolute error,root mean square error and mean absolute percentage error by 20.33%,36.79%and 10.12%respectively under the same type of faults,and the coefficient of determination can be improved by 0.11%.关键词
电缆智能诊断/优化格拉姆角场/混合卷积时序网络/特征提取Key words
intelligent diagnosis of cables/optimized Gram angle field/hybrid convolutional time-series networks/feature extraction分类
信息技术与安全科学引用本文复制引用
李练兵,代亮亮,高国强,王磊..基于VGAF与混合时序网络的电缆故障诊断方法[J].华中科技大学学报（自然科学版）,2025,53(7):16-23,8.基金项目
河北省省级科技计划资助项目(20312102D). （20312102D）
