基于TCN-BiLSTM网络的电力电缆故障诊断  

Fault Diagnosis of Mining Cable Based on TCN-BiLSTM Network

作　　者：胡业林[1] 王子涵 HU Yelin;WANG Zihan(College of Electrical and Information Engineering,Anhui University of Science and Technology,Huainan Anhui 232001,China)

机构地区：[1]安徽理工大学电气与信息工程学院,安徽淮南232001

出　　处：《佳木斯大学学报（自然科学版）》2024年第4期15-18,43,共5页Journal of Jiamusi University：Natural Science Edition

摘　　要：为了提升电力电缆故障诊断技术的准确率,解决传统电力电缆诊断过程中操作复杂、可靠性低和精准度不够等问题,提出了一种基于TCN和BiLSTM的电力电缆故障诊断方法。该方法的核心是使用Matlab/Simulink搭建三相电缆的仿真模型,按照电缆的实际参数设置模型,然后提取出电缆的四种短路故障:单相接地短路、双相接地短路、双相相间短路以及三相短路的电压信号。构建电缆故障样本集,搭建TCN和BiLSTM网络对电缆故障信号进行特征提取和序列捕捉,通过与TCN网络和CNN-BiLSTM网络进行实验对比,以及对从淮南某煤矿采集到的数据进行验证,证明该方法对电缆故障诊断具有良好的性能。In order to improve the accuracy of power cable fault diagnosis technology and solve the problems of complex operation,low reliability and insufficient accuracy in the traditional power cable diagnosis process,a power cable fault diagnosis method based on TCN and BiLSTM is proposed in this paper.The core of this method is to build a simulation model of three-phase cable using Matlab/Simulink,set the model according to the actual parameters of the cable,and then extract four kinds of short circuit faults of the cable:Voltage signals of single-phase ground short circuit,dual-phase ground short circuit,dual-phase short circuit and three-phase short circuit,cable fault sample set is constructed.TCN and BiLSTM networks are set up to carry out feature extraction and sequence capture of cable fault signals.Through experimental comparison with TCN network and CNN-BiLSTM network,the data collected from a coal mine in Huainan is verified,which proves that the method has good performance for cable fault diagnosis.

关 键 词：电缆 故障诊断 时域卷积网络 双向长短时记忆网络 短路故障 

分 类 号：TM247[一般工业技术—材料科学与工程]